{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.11.13","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"gpu","dataSources":[{"sourceId":81933,"databundleVersionId":9643020,"sourceType":"competition"}],"isInternetEnabled":false,"language":"python","sourceType":"notebook","isGpuEnabled":true}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"\n\n# Standard libraries\nimport warnings\nimport numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nimport seaborn as sns\nfrom collections import defaultdict\nfrom matplotlib.lines import Line2D\n\n# Scipy & stats\nfrom scipy.stats import chi2_contingency, mode\nfrom scipy.optimize import linear_sum_assignment\n\n# Sklearn - preprocessing\nfrom sklearn.model_selection import train_test_split, StratifiedKFold, GridSearchCV, RandomizedSearchCV\nfrom sklearn.preprocessing import StandardScaler\nfrom sklearn.impute import SimpleImputer\nfrom sklearn import clone\nfrom sklearn.base import BaseEstimator, TransformerMixin\n\n# Sklearn - models\nfrom sklearn.ensemble import RandomForestRegressor, RandomForestClassifier, IsolationForest\nfrom xgboost import XGBClassifier\nfrom sklearn.cluster import KMeans\n\n# Sklearn - feature selection\nfrom sklearn.feature_selection import SelectFromModel, RFECV\n\n\n\n# Sklearn - metrics\nfrom sklearn.metrics import (\n    accuracy_score, f1_score, precision_score, recall_score,\n    confusion_matrix, ConfusionMatrixDisplay,\n    cohen_kappa_score, roc_curve, auc, make_scorer\n)\n\n# Visualization helpers\nfrom kneed import KneeLocator\nimport umap\n\n# Sklearn - pipeline\nfrom imblearn.pipeline import Pipeline as ImbPipeline\nfrom imblearn.combine import SMOTEENN\n","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"### CARICAMENTO DEI DATI ###","metadata":{}},{"cell_type":"code","source":"def load_training_data():\n    return pd.read_csv('/kaggle/input/child-mind-institute-problematic-internet-use/train.csv')\n\ndef load_test_data():\n    return pd.read_csv('/kaggle/input/child-mind-institute-problematic-internet-use/test.csv')\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"training_data = load_training_data()\ntraining_data.describe()","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## Osservazioni sugli intervalli dei valori ##\nPossiamo osservare che i nostri dati presentano alcuni valori al di fuori dell'intervallo reale . Ad esempio, un peso pari a 0 o un valore di CGAS_Score pari a 999 indicano possibili anomalie nei dati. È importante sottolineare che i limiti minimi e massimi sono stati determinati attraverso ricerche online, senza una conoscenza diretta del dominio medico. Un’analisi condotta con il supporto di esperti del settore avrebbe potuto fornire una comprensione più accurata, riducendo il rischio di introdurre bias involontari nel modello che intendiamo sviluppare.","metadata":{}},{"cell_type":"code","source":"\n# voglio filtrare per range ma tenere i nan\n\ncondition = (training_data['CGAS-CGAS_Score'] >= 0) & (training_data['CGAS-CGAS_Score'] <= 100)\ntraining_data = training_data[condition | training_data['CGAS-CGAS_Score'].isna()]\n\ncondition = training_data['Physical-Weight'] > 0\ntraining_data = training_data[condition | training_data['Physical-Weight'].isna()]\n\n# Remove outliers for 'Fitness_Endurance-Max_Stage' but keep NaN\ncondition = training_data['Fitness_Endurance-Max_Stage'] > 0\ntraining_data = training_data[condition | training_data['Fitness_Endurance-Max_Stage'].isna()]\n\n# Remove outliers for 'Fitness_Endurance-Time_Mins' but keep NaN\ncondition = training_data['Fitness_Endurance-Time_Mins'] > 0\ntraining_data = training_data[condition | training_data['Fitness_Endurance-Time_Mins'].isna()]\n\ntraining_data[\"BIA-BIA_BMC\"] = np.where(training_data[\"BIA-BIA_BMC\"] <= 0, np.nan, training_data[\"BIA-BIA_BMC\"])\ntraining_data[\"BIA-BIA_BMC\"] = np.where(training_data[\"BIA-BIA_BMC\"] > 10, np.nan, training_data[\"BIA-BIA_BMC\"])\n# Remove outliers for 'BIA-BIA_DEE' but keep NaN\ncondition = training_data['BIA-BIA_DEE'] < 17311\ntraining_data = training_data[condition | training_data['BIA-BIA_DEE'].isna()]\n# Remove highly implausible values\n\n# Remove implausible body-fat\ntraining_data[\"BIA-BIA_Fat\"] = np.where(training_data[\"BIA-BIA_Fat\"] < 5, np.nan, training_data[\"BIA-BIA_Fat\"])\ntraining_data[\"BIA-BIA_Fat\"] = np.where(training_data[\"BIA-BIA_Fat\"] > 60, np.nan, training_data[\"BIA-BIA_Fat\"])\n# Basal Metabolic Rate\ntraining_data[\"BIA-BIA_BMR\"] = np.where(training_data[\"BIA-BIA_BMR\"] > 4000, np.nan, training_data[\"BIA-BIA_BMR\"])\n# Daily Energy Expenditure\ntraining_data[\"BIA-BIA_DEE\"] = np.where(training_data[\"BIA-BIA_DEE\"] > 8000, np.nan, training_data[\"BIA-BIA_DEE\"])\n# Fat Free Mass Index\ntraining_data[\"BIA-BIA_FFM\"] = np.where(training_data[\"BIA-BIA_FFM\"] <= 0, np.nan, training_data[\"BIA-BIA_FFM\"])\ntraining_data[\"BIA-BIA_FFM\"] = np.where(training_data[\"BIA-BIA_FFM\"] > 300, np.nan, training_data[\"BIA-BIA_FFM\"])\n# Fat Mass Index\ntraining_data[\"BIA-BIA_FMI\"] = np.where(training_data[\"BIA-BIA_FMI\"] < 0, np.nan, training_data[\"BIA-BIA_FMI\"])\n# Extra Cellular Water\ntraining_data[\"BIA-BIA_ECW\"] = np.where(training_data[\"BIA-BIA_ECW\"] > 100, np.nan, training_data[\"BIA-BIA_ECW\"])\n# Intra Cellular Water\ntraining_data[\"BIA-BIA_ICW\"] = np.where(training_data[\"BIA-BIA_ICW\"] > 100, np.nan, training_data[\"BIA-BIA_ICW\"])\n# Lean Dry Mass\ntraining_data[\"BIA-BIA_LDM\"] = np.where(training_data[\"BIA-BIA_LDM\"] > 100, np.nan, training_data[\"BIA-BIA_LDM\"])\n# Lean Soft Tissue\ntraining_data[\"BIA-BIA_LST\"] = np.where(training_data[\"BIA-BIA_LST\"] > 300, np.nan, training_data[\"BIA-BIA_LST\"])\n# Skeletal Muscle Mass\ntraining_data[\"BIA-BIA_SMM\"] = np.where(training_data[\"BIA-BIA_SMM\"] > 300, np.nan, training_data[\"BIA-BIA_SMM\"])\n# Total Body Water\ntraining_data[\"BIA-BIA_TBW\"] = np.where(training_data[\"BIA-BIA_TBW\"] > 300, np.nan, training_data[\"BIA-BIA_TBW\"])\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## Individuazione Problemi nel Dataset\nAndiamo ad osservare più in dettaglio in nostro dataset verificandone le potenziali problematiche:","metadata":{}},{"cell_type":"code","source":"# Conteggio delle righe totali\ntotale_righe = len(training_data)\n\n# Conteggio delle righe con almeno un valore nullo\nrighe_nulle = training_data.isnull().any(axis=1).sum()\n\n# Conteggio delle righe completamente non nulle\nrighe_non_nulle = totale_righe - righe_nulle\n\n# Creazione dell'istogramma\nplt.bar([\"Con almeno un valore nullo\", \"Non Nulle\", \"Totali\"], [righe_nulle, righe_non_nulle, totale_righe], color=['red', 'green', 'blue'])\nplt.xlabel(\"Dataset\")\nplt.ylabel(\"Conteggio\")\nplt.title(\"Conteggio di Righe Nulle, Non Nulle e Totali\")\nplt.show()","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Tutte le righe presentano un dato almeno non valido, analizziamo in modo più approfondito la distribuzione dei valori Nan","metadata":{}},{"cell_type":"code","source":"null_counts = training_data.isnull().sum()\nprint(\" \\nCount total NaN at each column in a DataFrame : \\n\\n\",null_counts)\nprint(\" \\nRate total NaN at each column in a DataFrame : \\n\\n\",null_counts / len(training_data))\nplt.figure(figsize=(18, 6))\nplt.bar(null_counts.index, null_counts.values/len(training_data), color='teal')\nplt.xlabel('Features', fontsize=12)\nplt.ylabel('Null Value Count', fontsize=12)\nplt.title('Count of Null Values by Feature', fontsize=14)\nplt.xticks(rotation=90, fontsize=10)\nplt.tight_layout()\nplt.show()","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Il dataset contiente una presenza non trascurabile di valori mancanti in quasi tutte le colonne. La colonna riguardante la variabile target Sii presenta anch'essa un 30% dei dati non validi, una imputazione su questa colonna potrebbe provocare un esplosione dell'errore sulle previsioni. Possiamo provare ad imputare alcuni valori mancanti di Sii derivandolo dalle risposte alle domdande date.","metadata":{}},{"cell_type":"code","source":"totale_sii_missing = training_data['sii'].isna().sum()\ntotale_missing_target = training_data[['sii', 'PCIAT-PCIAT_01', 'PCIAT-PCIAT_02', 'PCIAT-PCIAT_03', 'PCIAT-PCIAT_04',\n                                       'PCIAT-PCIAT_05', 'PCIAT-PCIAT_06', 'PCIAT-PCIAT_07', 'PCIAT-PCIAT_08',\n                                       'PCIAT-PCIAT_09', 'PCIAT-PCIAT_10', 'PCIAT-PCIAT_11', 'PCIAT-PCIAT_12',\n                                       'PCIAT-PCIAT_13', 'PCIAT-PCIAT_14', 'PCIAT-PCIAT_15', 'PCIAT-PCIAT_16',\n                                       'PCIAT-PCIAT_17', 'PCIAT-PCIAT_18', 'PCIAT-PCIAT_19', 'PCIAT-PCIAT_20','PCIAT-PCIAT_Total']].isna().all(axis=1).sum()\n# Creazione dell'istogramma\nplt.bar([\"Sii NAN\", \"Sii+ colonne dipendenti\"], [totale_sii_missing, totale_missing_target], color=['red', 'green'])\nplt.xlabel(\"Dataset\")\nplt.title(\"Conteggio di Righe Sii Nulli, Sii + risposte alle domande\")\nplt.ylabel(\"Conteggio\")\n\nplt.show()","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"I dati presentano degli sii non validi ( Nan ). Una strategia che possiamo adottare è quella di ricostruirci 'sii' partendo dalle risposte che sono state fornite dal paziente.","metadata":{}},{"cell_type":"markdown","source":"","metadata":{}},{"cell_type":"code","source":"\n# recupero tutti i sii missing\n# conto quante risposte sono missing\n\n# Conta i valori mancanti per ciascuna colonna\n\nansw_columns = [\n    'sii',\n    'PCIAT-PCIAT_01', 'PCIAT-PCIAT_02', 'PCIAT-PCIAT_03', 'PCIAT-PCIAT_04',\n    'PCIAT-PCIAT_05', 'PCIAT-PCIAT_06', 'PCIAT-PCIAT_07', 'PCIAT-PCIAT_08',\n    'PCIAT-PCIAT_09', 'PCIAT-PCIAT_10', 'PCIAT-PCIAT_11', 'PCIAT-PCIAT_12',\n    'PCIAT-PCIAT_13', 'PCIAT-PCIAT_14', 'PCIAT-PCIAT_15', 'PCIAT-PCIAT_16',\n    'PCIAT-PCIAT_17', 'PCIAT-PCIAT_18', 'PCIAT-PCIAT_19', 'PCIAT-PCIAT_20',\n    'PCIAT-PCIAT_Total'\n]\n\n\nmissing_counts = training_data[answ_columns].isna().sum()\n\nplt.figure(figsize=(12,6))\nmissing_counts.plot(kind='bar', color='skyblue', edgecolor='black')\n\nplt.title(\"Valori mancanti per colonna\", fontsize=14)\nplt.xlabel(\"Colonne\", fontsize=12)\nplt.ylabel(\"Numero di missing\", fontsize=12)\nplt.xticks(rotation=45, ha='right')\nplt.tight_layout()\nplt.show()\n\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"La strategia di assegnamento di valori di 'sii' in base alle resposte parziali del paziente non risulta valida poiché tutti gli sii mancanti sono associati anche a tutte risposte non date.\n\nAndremo a rimuovere tutte le righe con 'sii' a NaN per evitare di introdurre bias alla classificazione di classi rumorose date dalla imputazione.","metadata":{}},{"cell_type":"code","source":"training_data = training_data[training_data['sii'].notna()]","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Le colonne PAQ_A-PAQ_A_Total e PAQ_C-PAQ_C_Total sono i relativi punteggi di attività fisica attruibuiti da professionisti nel caso sia un adolescente o nel caso sia un bambino. Queste due colonne hanno lo stesso significato ma l'importanza cambia in scala all'età che stiamo prendendo in considerazione. Possiamo unire queste due colonne facendo in modo di tenere anche l'età come valore per non perdere il significato della feature.","metadata":{}},{"cell_type":"code","source":"training_data['PAQ_TOTAL'] = training_data['PAQ_A-PAQ_A_Total'].fillna(training_data['PAQ_C-PAQ_C_Total'])\ntraining_data.drop(inplace=True,axis=1,columns=\"PAQ_A-PAQ_A_Total\")\ntraining_data.drop(inplace=True,axis=1,columns=\"PAQ_C-PAQ_C_Total\")","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Si è scelto di eliminare le colonne con una proprorzione superiore al 49% di missing values per evitare di non introdurre bias nel modello predittivo. Il bias è dato dall'elevata imputazione delle colonne.\n\nUn altro motivo della eliminazione di colonne è che l'imputazione non sarebbe utile alla classificazione perchè le righe possiedono pressochè lo stesso valore ( dovuto al fatto che più di metà sono mancanti ).","metadata":{}},{"cell_type":"code","source":"thresh = 0.49 * len(training_data)\ntraining_data.dropna(thresh = thresh, axis = 1, inplace = True)\nnull_counts = training_data.isnull().sum()\nprint(\" \\nCount total NaN at each column in a DataFrame : \\n\\n\",null_counts)\nprint(\" \\nRate total NaN at each column in a DataFrame : \\n\\n\",null_counts / len(training_data))\nplt.figure(figsize=(18, 6))\nplt.bar(null_counts.index, null_counts.values/len(training_data), color='teal')\nplt.xlabel('Features', fontsize=12)\nplt.ylabel('Null Value Count', fontsize=12)\nplt.title('Count of Null Values by Feature', fontsize=14)\nplt.xticks(rotation=90, fontsize=10)\nplt.tight_layout()\nplt.show()","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Le osservazioni che contengono poca informazione, ovvero quelle con una percentuale di valori nulli molto elevata,vengono eliminate poiché necessaria un'imputazione massiva sulla riga.","metadata":{}},{"cell_type":"code","source":"num_cols = len(training_data.columns)\n\n# Definiamo le soglie da analizzare (dal 10% al 100%)\nsoglie = np.arange(0.1, 1.1, 0.1)\n\n# Lista per salvare il numero di righe per ogni soglia\nrighe_per_soglia = []\n\n# Calcola il numero di righe che superano ogni soglia di NaN\nfor soglia in soglie:\n    num_nan = int(soglia * num_cols)\n    righe_con_nan = (training_data.isna().sum(axis=1) >= num_nan).sum()\n    righe_per_soglia.append(righe_con_nan)\n\nplt.figure(figsize=(10, 5))\nplt.bar([f\"{int(s*100)}%\" for s in soglie], righe_per_soglia, color=\"red\")\n\n# Aggiunta etichette e titolo\nplt.xlabel(\"Percentuale di valori NaN nella riga\")\nplt.ylabel(\"Conteggio delle righe\")\nplt.title(\"Distribuzione del numero di righe per proporzione di valori mancanti\")\nplt.xticks(rotation=45)\nplt.show()","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Le righe totali al massimo presentano una mancanza del 60% dei dati, con una distribuzione elevata per osservazioni sopra al 50%.\n\n## Azioni preliminari ##\n# Rimozione di variabili rindondanti: #\n- BMI = peso (kg) / (statura (m) x statura (m)) è l'indice di massa corporea. Il BMI è ottenuto dai valori nelle colonne Height e Weight. Per questo si decide di tenere solo come valore il BMI andando ad eliminare le altre due colonne.\n- Physical-BMI e BIA-BIA_BMI rappresentano la stessa misura ma Physical-BMI ha mediamente una precisione migliore perchè derivante da una Bia professionale a discapito di BIA-BIA_BMI che è registrata con una bilancia media. Per questo motivo, si predilige il valore Physical-BMI se presente. \n","metadata":{}},{"cell_type":"code","source":"training_data.drop(columns=['Physical-Height','Physical-Weight'],inplace=True,axis=1,errors='ignore')\n\ndef select_bmi(row):\n    if pd.notna(row['BIA-BIA_BMI']):\n        return row['BIA-BIA_BMI']\n    else:\n        return row['Physical-BMI']\n\ntraining_data['BMI'] = training_data.apply(select_bmi, axis=1)\ntraining_data.drop(columns=['BIA-BIA_BMI','Physical-BMI'],inplace=True,axis=1,errors='ignore')","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Dividiamo le variabili di test con quelle di visualizzabili solo nel training.","metadata":{}},{"cell_type":"code","source":"target_columns = ['PCIAT-PCIAT_01', 'PCIAT-PCIAT_02', 'PCIAT-PCIAT_03', 'PCIAT-PCIAT_04', 'PCIAT-PCIAT_05', 'PCIAT-PCIAT_06', 'PCIAT-PCIAT_07', 'PCIAT-PCIAT_08', 'PCIAT-PCIAT_09', 'PCIAT-PCIAT_10', 'PCIAT-PCIAT_11', 'PCIAT-PCIAT_12', 'PCIAT-PCIAT_13', 'PCIAT-PCIAT_14', 'PCIAT-PCIAT_15', 'PCIAT-PCIAT_16', 'PCIAT-PCIAT_17', 'PCIAT-PCIAT_18', 'PCIAT-PCIAT_19', 'PCIAT-PCIAT_20', 'PCIAT-PCIAT_Total','sii']\n\ncategorical_columns = ['Basic_Demos-Enroll_Season',\n                       'CGAS-Season',\n                       'Physical-Season',\n                       'Fitness_Endurance-Season',\n                       'FGC-Season',\n                       'BIA-Season',\n                       'PAQ_A-Season',\n                       'PAQ_C-Season',\n                       'PCIAT-Season',\n                       'PreInt_EduHx-Season',\n                       'Basic_Demos-Enroll_Season_Summer',\n                       'Basic_Demos-Enroll_Season_Winter', 'CGAS-Season_Spring',\n                       'CGAS-Season_Summer', 'CGAS-Season_Winter', 'Physical-Season_Spring',\n                       'Physical-Season_Summer', 'Physical-Season_Winter',\n                       'Fitness_Endurance-Season_Spring', 'Fitness_Endurance-Season_Summer',\n                       'Fitness_Endurance-Season_Winter', 'FGC-Season_Spring',\n                       'FGC-Season_Summer', 'FGC-Season_Winter', 'BIA-Season_Spring',\n                       'BIA-Season_Summer', 'BIA-Season_Winter', 'PAQ_A-Season_Spring',\n                       'PAQ_A-Season_Summer', 'PAQ_A-Season_Winter', 'PAQ_C-Season_Spring',\n                       'PAQ_C-Season_Summer', 'PAQ_C-Season_Winter', 'PCIAT-Season_Spring',\n                       'PCIAT-Season_Summer', 'PCIAT-Season_Winter','Basic_Demos-Enroll_Season_Spring',\n                       'SDS-Season','PreInt_EduHx-Season_Spring', 'PreInt_EduHx-Season_Summer',\n                       'PreInt_EduHx-Season_Winter', 'SDS-Season_Spring', 'SDS-Season_Summer',\n                       'SDS-Season_Winter','Basic_Demos-Enroll_Season_Spring','Basic_Demos-Sex','FGC-FGC_PU_Zone','FGC-FGC_SRL_Zone','FGC-FGC_CU_Zone','sii','PAQ_C-Season_Fall' ,'FGC-FGC_TL_Zone' , 'FGC-FGC_SRR_Zone']\n\n\ndef split_X_Y(dataset: pd.DataFrame, target_column: str, test_size: float = 0.25, random_state: int = 42):\n    X = dataset.drop(columns=[target_column])\n    Y = dataset[target_column]\n\n    X_train, X_test, Y_train, Y_test = train_test_split(\n        X, Y, test_size=test_size, random_state=random_state, stratify=Y\n    )\n\n    return X_train, X_test, Y_train, Y_test\n\n\nX_train,X_test,Y_train,Y_test = split_X_Y(training_data,target_column=\"sii\")\n# remove id from train\n\nX_train = X_train.drop(columns=\"id\")\nX_test = X_test.drop(columns=\"id\")\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## Distribuzuione delle classi Sii ##","metadata":{}},{"cell_type":"code","source":"plt.hist(training_data['sii'])\nplt.title(\"Sii Distribuition\")","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"La classe 'sii' presenta un sbilanciamento proporzionale al numero di classe. Siamo in presenza di numerose osservazioni della classe 0 e pochissime osservazioni di classe 3.","metadata":{}},{"cell_type":"markdown","source":"### GESTIONE DELLE VARIABILI CATEGORIALI NON ORDINALI ###\n\nSiamo in presenza di numerose feature di classi perchè sklearn.RandomForest non permette l'utilizzo di feature codificate come stringhe. Infatti, siamo costretti ad effettuare una one-hot encoding così da permettere di trasformare una variabile codificata come stringa in una variabile binaria. Il contro dell'utilizzo dell'one-hot encoding è l'aumento del numero di dimensioni del nostro dataset che porta ad un peggioramento delle prestazioni degli algoritmi e a favorire la curse of dimensionality.\n\nPer questo motivo verranno selezionate solo le classi significativamente utili alla classificazione della variabile 'sii'. Per fare ciò si utilizzano delle tabelle di contingenza e si applica il test del chi-quadro selezionando solo le variabili significative.","metadata":{}},{"cell_type":"code","source":"season_columns = [col for col in X_train.columns if \"Season\" in col in col]\n\n# Inizializza lista per salvare i risultati\nresults = []\n\n# Calcola chi-quadro per ciascuna variabile stagionale\nfor col in season_columns:\n    try:\n        table = pd.crosstab(X_train[col], Y_train)\n        chi2, p, _, _ = chi2_contingency(table)\n        results.append({'column': col, 'p_value': p})\n    except Exception as e:\n        results.append({'column': col, 'p_value': None, 'error': str(e)})\n\n# Crea il DataFrame dei risultati\nchi2_df = pd.DataFrame(results).sort_values(by='p_value')\nchi2_df\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"La variabile PAQ_C-Season ha un p-value di 0.022 per il quale si può accettare l'ipotesi che abbia una significatità nella classificazione del 'sii'","metadata":{}},{"cell_type":"code","source":"X_train_encoded = pd.get_dummies(X_train, columns=['PAQ_C-Season'])\nX_test_encoded = pd.get_dummies(X_test, columns=['PAQ_C-Season'])\n#rimuovo tutte le altre\n\nX_train_encoded = X_train_encoded.drop(columns=[col[\"column\"] for col in results if col[\"p_value\"]> 0.02])\nX_test_encoded = X_test_encoded.drop(columns=[col[\"column\"] for col in results if col[\"p_value\"]> 0.02])\nX_train_encoded.describe()","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Le classi categoriche verranno sostuite con la classe più frequente ( moda ).\n\nPer le variabili continue invece:\n\nattraverso media ( metodi di imputazione Semplici )\nattraverso una combinazione delle altre features ( metodi di imputazione Iterativi )\nnon facendo imputazione\nNB: per motivi di velocità nella computazione verrà solo affrontata l'imputazione attraverso la media in questo notebook","metadata":{}},{"cell_type":"code","source":"def fill_data(dataset: pd.DataFrame, categorical_columns=None, method=\"median\",iterative_imputer=RandomForestRegressor()):\n    \"\"\"\n    Imputa i valori mancanti in un DataFrame, separando dati categorici e continui.\n    \"\"\"\n\n    categorical_columns = [col for col in dataset.columns if col in categorical_columns]\n\n    if categorical_columns is None:\n        categorical_columns = []\n    else:\n        categorical_columns = [col for col in dataset.columns if col in categorical_columns]\n\n\n    continuous_columns = [col for col in dataset.columns if col not in categorical_columns]\n\n    filled_data = dataset.copy()\n\n    imputers = {}\n\n    # Imputazione categorica con SimpleImputer\n    if categorical_columns:\n        cat_imputer = SimpleImputer(strategy='most_frequent')\n        filled_data[categorical_columns] = cat_imputer.fit_transform(filled_data[categorical_columns])\n        imputers['categorical'] = cat_imputer\n\n    # Imputazione continua\n    if method in [\"mean\"]:\n        cont_imputer = SimpleImputer(strategy=method)\n    else:\n        #cont_imputer = IterativeImputer(estimator=iterative_imputer, max_iter=10, random_state=0)\n        None\n        \n    if continuous_columns:\n        filled_data[continuous_columns] = cont_imputer.fit_transform(filled_data[continuous_columns])\n        imputers['continuous'] = cont_imputer\n\n    return filled_data, imputers\n\n\ndef apply_imputers(X_test: pd.DataFrame, imputers: dict, categorical_columns: list[str]) -> pd.DataFrame:\n    \"\"\"\n    Applica imputers già fittati a X_test.\n    \"\"\"\n    filled_test = X_test.copy()\n\n    categorical_columns = [col for col in X_test.columns if col in categorical_columns]\n    continuous_columns = [col for col in X_test.columns if col not in categorical_columns]\n\n    if 'categorical' in imputers and categorical_columns:\n        filled_test[categorical_columns] = imputers['categorical'].transform(filled_test[categorical_columns])\n\n    if 'continuous' in imputers and continuous_columns:\n        filled_test[continuous_columns] = imputers['continuous'].transform(filled_test[continuous_columns])\n\n    return filled_test\n\n\nX_train_not_imputed = X_train_encoded.dropna()\nY_train_not_imputed = Y_train.loc[X_train_not_imputed.index]\n\nX_train_imputed_mean,imputer_mean =  fill_data(dataset=X_train_encoded,categorical_columns=categorical_columns,method=\"mean\")\n#X_train_imputed_c,imputer_iterative =  fill_data(dataset=X_train_encoded,categorical_columns=categorical_columns,method=\"iterative\")\n\n\nX_test_not_imputed = X_test_encoded.dropna()\nY_test_not_imputed = Y_test.loc[X_test_not_imputed.index]\n\nX_test_imputed_mean =  apply_imputers(X_test=X_test_encoded,imputers=imputer_mean,categorical_columns=categorical_columns)\n#X_test_iterative =  apply_imputers(X_test=X_train_encoded,imputers=imputer_iterative,categorical_columns=categorical_columns)\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"### TECNICHE PER BILANCIAMENTO DELLE CLASSI ###\n\nI metodi per il bilanciamento delle classi sono SMOTEENN e il bilanciamento dei pesi nel modello di random forest. In seguito verrà fatta una comparazione tra i due algoritmi creati per valutarne la bontà del modello.\n\n## Funzionamento SMOTEENN ##\nSMOTEENN è un algoritmo di over-sampling e di under-sampling dei dati in presenza di dati sbilanciati. Si divide in due operazioni fondamentali:\n\n- SMOTE: va a generare dei dati sintetici utilizzando la media dei K dati più vicini, metodo di oversampling.\n- EEN: edited nearest neighbors, metodo di undersampling, rimuove i dati rumorosi attraverso KNN ( utilizzato un KNN con un default k = 3, conteggiati solo 3 vicini ), vengono rimosse le classi minoritarie presenti nel gruppo del k-vicinato.\n  \nContro:\n\n- stiamo introducendo nuovi iperparametri da tunare con l'inserimento di EEN.\n","metadata":{}},{"cell_type":"code","source":"def standardizer(dataset):\n    scaler = StandardScaler()\n    dataset = scaler.fit_transform(dataset)\n    return dataset,scaler\n\ndef apply_standardizer(standardizer:StandardScaler,X_test):\n    return standardizer.transform(X_test)\n\nX_train_not_imputed_std, standardizer_default = standardizer(X_train_not_imputed)\nX_train_imputed_mean_std, standardizer_mean = standardizer(X_train_imputed_mean)\n#X_train_imputed_c_std, standardizer_c = standardizer(X_train_imputed_c)\n\nX_test_not_imputed_std = apply_standardizer(standardizer_default,X_test_not_imputed)\nX_test_imputed_mean_std = apply_standardizer(standardizer=standardizer_mean,X_test=X_test_imputed_mean)\n#X_test_imputed_c_std = apply_standardizer(standardizer=standardizer_c, X_test_iterative)","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"smoteenn = SMOTEENN(random_state=42)\n\nX_train_not_imputed_resampled, Y_train_not_imputed_resampled = smoteenn.fit_resample(X_train_not_imputed_std, Y_train_not_imputed)\n\nX_train_imputed_mean_resampled, Y_train_mean_resampled = smoteenn.fit_resample(X_train_imputed_mean_std, Y_train)\n#X_train_imputed_c_resampled, Y_train_c_resampled = smoteenn.fit_resample(X_train_imputed_c_std, Y_train)\n\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"### Script Contentente tutto quello illustrato fino ad ora ###\nVengono addestrati modelli su un K-Fold Cross-Validation, i modelli testati sono: \n\n- Random Forest\n- XGBoost\n\nSeguiranno la seguente pipeline:\n\n![pipeline.jpg](attachment:560c7e35-56e1-4e9d-a22e-13bb4f46af1a.jpg)\n\nI modelli veranno tunati attraverso una random-Search su una griglia di iperparametri prestabilita, la performance dei modelli dipende dalla selezione preliminare di questi parametri, una miglior selezione potrebbe portare ad un'affidabilità maggiore dei modelli.\n\nVerranno testate tutte le possibili pipeline:\n\n- No RFECV + SMOTENN\n- RFECV + SMOTENN\n- No RFECV + NO SMOTENN\n- RFECV + SMOTENN\n\nLa predizione del modello sarà data dalla maggioranza dei voti stabiliti dal gruppo di modelli costruiti all'interno della pipeline","metadata":{},"attachments":{"560c7e35-56e1-4e9d-a22e-13bb4f46af1a.jpg":{"image/jpeg":"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"}}},{"cell_type":"code","source":"def load_clean_training_data():\n    training_data = pd.read_csv('/kaggle/input/child-mind-institute-problematic-internet-use/train.csv')\n    condition = (training_data['CGAS-CGAS_Score'] >= 0) & (training_data['CGAS-CGAS_Score'] <= 100)\n    training_data = training_data[condition | training_data['CGAS-CGAS_Score'].isna()]\n\n    condition = training_data['Physical-Weight'] > 0\n    training_data = training_data[condition | training_data['Physical-Weight'].isna()]\n\n    # Remove outliers for 'Fitness_Endurance-Max_Stage' but keep NaN\n    condition = training_data['Fitness_Endurance-Max_Stage'] > 0\n    training_data = training_data[condition | training_data['Fitness_Endurance-Max_Stage'].isna()]\n\n    # Remove outliers for 'Fitness_Endurance-Time_Mins' but keep NaN\n    condition = training_data['Fitness_Endurance-Time_Mins'] > 0\n    training_data = training_data[condition | training_data['Fitness_Endurance-Time_Mins'].isna()]\n\n    training_data[\"BIA-BIA_BMC\"] = np.where(training_data[\"BIA-BIA_BMC\"] <= 0, np.nan, training_data[\"BIA-BIA_BMC\"])\n    training_data[\"BIA-BIA_BMC\"] = np.where(training_data[\"BIA-BIA_BMC\"] > 10, np.nan, training_data[\"BIA-BIA_BMC\"])\n    # Remove outliers for 'BIA-BIA_DEE' but keep NaN\n    condition = training_data['BIA-BIA_DEE'] < 17311\n    training_data = training_data[condition | training_data['BIA-BIA_DEE'].isna()]\n    # Remove highly implausible values\n\n    # Remove implausible body-fat\n    training_data[\"BIA-BIA_Fat\"] = np.where(training_data[\"BIA-BIA_Fat\"] < 5, np.nan, training_data[\"BIA-BIA_Fat\"])\n    training_data[\"BIA-BIA_Fat\"] = np.where(training_data[\"BIA-BIA_Fat\"] > 60, np.nan, training_data[\"BIA-BIA_Fat\"])\n    # Basal Metabolic Rate\n    training_data[\"BIA-BIA_BMR\"] = np.where(training_data[\"BIA-BIA_BMR\"] > 4000, np.nan, training_data[\"BIA-BIA_BMR\"])\n    # Daily Energy Expenditure\n    training_data[\"BIA-BIA_DEE\"] = np.where(training_data[\"BIA-BIA_DEE\"] > 8000, np.nan, training_data[\"BIA-BIA_DEE\"])\n    # Fat Free Mass Index\n    training_data[\"BIA-BIA_FFM\"] = np.where(training_data[\"BIA-BIA_FFM\"] <= 0, np.nan, training_data[\"BIA-BIA_FFM\"])\n    training_data[\"BIA-BIA_FFM\"] = np.where(training_data[\"BIA-BIA_FFM\"] > 300, np.nan, training_data[\"BIA-BIA_FFM\"])\n    # Fat Mass Index\n    training_data[\"BIA-BIA_FMI\"] = np.where(training_data[\"BIA-BIA_FMI\"] < 0, np.nan, training_data[\"BIA-BIA_FMI\"])\n    # Extra Cellular Water\n    training_data[\"BIA-BIA_ECW\"] = np.where(training_data[\"BIA-BIA_ECW\"] > 100, np.nan, training_data[\"BIA-BIA_ECW\"])\n    # Intra Cellular Water\n    training_data[\"BIA-BIA_ICW\"] = np.where(training_data[\"BIA-BIA_ICW\"] > 100, np.nan, training_data[\"BIA-BIA_ICW\"])\n    # Lean Dry Mass\n    training_data[\"BIA-BIA_LDM\"] = np.where(training_data[\"BIA-BIA_LDM\"] > 100, np.nan, training_data[\"BIA-BIA_LDM\"])\n    # Lean Soft Tissue\n    training_data[\"BIA-BIA_LST\"] = np.where(training_data[\"BIA-BIA_LST\"] > 300, np.nan, training_data[\"BIA-BIA_LST\"])\n    # Skeletal Muscle Mass\n    training_data[\"BIA-BIA_SMM\"] = np.where(training_data[\"BIA-BIA_SMM\"] > 300, np.nan, training_data[\"BIA-BIA_SMM\"])\n    # Total Body Water\n    training_data[\"BIA-BIA_TBW\"] = np.where(training_data[\"BIA-BIA_TBW\"] > 300, np.nan, training_data[\"BIA-BIA_TBW\"])\n\n    training_data = training_data[training_data['sii'].notna()]\n\n    training_data['PAQ_TOTAL'] = training_data['PAQ_A-PAQ_A_Total'].fillna(training_data['PAQ_C-PAQ_C_Total'])\n    training_data.drop(inplace=True,axis=1,columns=\"PAQ_A-PAQ_A_Total\")\n    training_data.drop(inplace=True,axis=1,columns=\"PAQ_C-PAQ_C_Total\")\n\n    thresh = 0.49 * len(training_data)\n    training_data.dropna(thresh = thresh, axis = 1, inplace = True)\n    training_data.drop(columns=['Physical-Height','Physical-Weight'],inplace=True,axis=1,errors='ignore')\n\n    def select_bmi(row):\n        if pd.notna(row['BIA-BIA_BMI']):\n            return row['BIA-BIA_BMI']\n        else:\n            return row['Physical-BMI']\n\n    training_data['BMI'] = training_data.apply(select_bmi, axis=1)\n    training_data.drop(columns=['BIA-BIA_BMI','Physical-BMI'],inplace=True,axis=1,errors='ignore')\n\n    training_data = training_data.drop(columns=[\"SDS-Season\",\"FGC-Season\",\"Basic_Demos-Enroll_Season\",\"Physical-Season\",\"PreInt_EduHx-Season\",\"BIA-Season\",\n                                                    \"PCIAT-Season\",\"CGAS-Season\"],errors='ignore')\n\n    training_data =  training_data.drop(columns=[\"SDS-Season\",\"FGC-Season\",\"Basic_Demos-Enroll_Season\",\"Physical-Season\",\"PreInt_EduHx-Season\",\"BIA-Season\",\n                                                   \"PCIAT-Season\",\"CGAS-Season\"],errors='ignore')\n\n    training_data = pd.get_dummies(training_data, columns=['PAQ_C-Season'])\n\n    return training_data\n\n\ndef load_clean_test_data():\n    test_data_ = pd.read_csv('/kaggle/input/child-mind-institute-problematic-internet-use/test.csv')\n\n    test_data_['PAQ_TOTAL'] = test_data_['PAQ_A-PAQ_A_Total'].fillna(test_data_['PAQ_C-PAQ_C_Total'])\n    test_data_.drop(inplace=True,axis=1,columns=\"PAQ_A-PAQ_A_Total\")\n    test_data_.drop(inplace=True,axis=1,columns=\"PAQ_C-PAQ_C_Total\")\n    test_data_.drop(columns=['Physical-Height','Physical-Weight'],inplace=True,axis=1,errors='ignore')\n\n    def select_bmi(row):\n\n        if pd.notna(row['BIA-BIA_BMI']):\n            return row['BIA-BIA_BMI']\n        else:\n            return row['Physical-BMI']\n\n    test_data_['BMI'] = test_data_.apply(select_bmi, axis=1)\n    test_data_.drop(columns=['BIA-BIA_BMI','Physical-BMI'],inplace=True,axis=1,errors='ignore')\n\n    test_data_ = test_data_.drop(columns=[\"SDS-Season\",\"FGC-Season\",\"Basic_Demos-Enroll_Season\",\"Physical-Season\",\"PreInt_EduHx-Season\",\"BIA-Season\",\n                                                \"PCIAT-Season\",\"CGAS-Season\"],errors='ignore')\n\n    test_data_ = pd.get_dummies(test_data_, columns=['PAQ_C-Season'])\n\n    return test_data_\n\ntarget_columns = ['PCIAT-PCIAT_01', 'PCIAT-PCIAT_02', 'PCIAT-PCIAT_03', 'PCIAT-PCIAT_04', 'PCIAT-PCIAT_05', 'PCIAT-PCIAT_06', 'PCIAT-PCIAT_07', 'PCIAT-PCIAT_08', 'PCIAT-PCIAT_09', 'PCIAT-PCIAT_10', 'PCIAT-PCIAT_11', 'PCIAT-PCIAT_12', 'PCIAT-PCIAT_13', 'PCIAT-PCIAT_14', 'PCIAT-PCIAT_15', 'PCIAT-PCIAT_16', 'PCIAT-PCIAT_17', 'PCIAT-PCIAT_18', 'PCIAT-PCIAT_19', 'PCIAT-PCIAT_20', 'PCIAT-PCIAT_Total','sii']\n\ncategorical_columns = ['Basic_Demos-Enroll_Season',\n                       'CGAS-Season',\n                       'Physical-Season',\n                       'Fitness_Endurance-Season',\n                       'FGC-Season',\n                       'BIA-Season',\n                       'PAQ_A-Season',\n                       'PAQ_C-Season',\n                       'PCIAT-Season',\n                       'PreInt_EduHx-Season',\n                       'Basic_Demos-Enroll_Season_Summer',\n                       'Basic_Demos-Enroll_Season_Winter', 'CGAS-Season_Spring',\n                       'CGAS-Season_Summer', 'CGAS-Season_Winter', 'Physical-Season_Spring',\n                       'Physical-Season_Summer', 'Physical-Season_Winter',\n                       'Fitness_Endurance-Season_Spring', 'Fitness_Endurance-Season_Summer',\n                       'Fitness_Endurance-Season_Winter', 'FGC-Season_Spring',\n                       'FGC-Season_Summer', 'FGC-Season_Winter', 'BIA-Season_Spring',\n                       'BIA-Season_Summer', 'BIA-Season_Winter', 'PAQ_A-Season_Spring',\n                       'PAQ_A-Season_Summer', 'PAQ_A-Season_Winter', 'PAQ_C-Season_Spring',\n                       'PAQ_C-Season_Summer', 'PAQ_C-Season_Winter', 'PCIAT-Season_Spring',\n                       'PCIAT-Season_Summer', 'PCIAT-Season_Winter','Basic_Demos-Enroll_Season_Spring',\n                       'SDS-Season','PreInt_EduHx-Season_Spring', 'PreInt_EduHx-Season_Summer',\n                       'PreInt_EduHx-Season_Winter', 'SDS-Season_Spring', 'SDS-Season_Summer',\n                       'SDS-Season_Winter','Basic_Demos-Enroll_Season_Spring','Basic_Demos-Sex','FGC-FGC_PU_Zone','FGC-FGC_SRL_Zone','FGC-FGC_CU_Zone']\n\n\nleaky_features = [\n    'PCIAT-PCIAT_01', 'PCIAT-PCIAT_02', 'PCIAT-PCIAT_03', 'PCIAT-PCIAT_04',\n    'PCIAT-PCIAT_05', 'PCIAT-PCIAT_06', 'PCIAT-PCIAT_07', 'PCIAT-PCIAT_08',\n    'PCIAT-PCIAT_09', 'PCIAT-PCIAT_10', 'PCIAT-PCIAT_11', 'PCIAT-PCIAT_12',\n    'PCIAT-PCIAT_13', 'PCIAT-PCIAT_14', 'PCIAT-PCIAT_15', 'PCIAT-PCIAT_16',\n    'PCIAT-PCIAT_17', 'PCIAT-PCIAT_18', 'PCIAT-PCIAT_19', 'PCIAT-PCIAT_20',\n    'PCIAT-PCIAT_Total'\n]\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## Creazione di una classe per gestire la pipeline ##","metadata":{}},{"cell_type":"code","source":"class ModelComparisonPipeline:\n    def __init__(self, model_sklearn, model_external,\n                 param_grid_sklearn=None, param_grid_external=None,\n                 external_search_type=\"random\", n_iter_external=30,\n                 n_splits=5,hyper_cv = 5,rf_cv = 5):\n        self.model_sklearn = model_sklearn\n        self.model_external = model_external\n        self.param_grid_sklearn = param_grid_sklearn\n        self.param_grid_external = param_grid_external\n        self.external_search_type = external_search_type\n        self.n_iter_external = n_iter_external\n        self.n_splits = n_splits\n        self.hyper_cv = hyper_cv\n        self.rf_cv = rf_cv\n        self.use_resampling = True\n        self.results = defaultdict(lambda: defaultdict(list))\n        self.external_group = []\n        self.sklearn_group = []\n        self.feature_columns_sklearn = []\n        self.feature_columns_external = []\n        \n\n    def apply_RFECV(self,sklearn_model,external_model,X,y,**kwargs):\n        \n        sklearn_pipeline = self._make_pipeline(sklearn_model)\n        external_pipeline = self._make_pipeline(external_model)\n        \n        cv_for_rfe = StratifiedKFold(n_splits=self.rf_cv, shuffle=True, random_state=42)\n        \n        selector_sklearn = RFECV(\n            estimator=sklearn_pipeline,\n            step=1,\n            scoring=make_scorer(cohen_kappa_score, weights='quadratic'),\n            cv=cv_for_rfe,\n            min_features_to_select=1,\n            n_jobs=-1,\n            importance_getter='named_steps.model.feature_importances_'\n        )\n\n        selector_external = RFECV(\n            estimator=external_pipeline,\n            step=1,\n            scoring=make_scorer(cohen_kappa_score, weights='quadratic'),\n            cv=cv_for_rfe,\n            min_features_to_select=1,\n            n_jobs=-1,\n            importance_getter='named_steps.model.feature_importances_'\n        )\n\n\n        imputer = SimpleImputer(strategy=\"mean\") \n        X_imputed = imputer.fit_transform(X)\n        \n        selector_sklearn.fit(X_imputed,y)\n        selector_external.fit(X_imputed,y)\n\n        return X.columns[selector_sklearn.support_],X.columns[selector_external.support_]\n\n\n    def start_pipeline(self, df,\n                       imputation_strategy='mean',\n                       use_resampling=True,\n                       use_RFE=True,**kwargs):\n        \"\"\"\n            Executes training/evaluation with Stratified K-Fold CV.\n        \"\"\"\n        X = df.drop(columns=['sii'])\n        y = df['sii']\n\n        all_columns = X.columns\n\n        X_sklearn = X\n        X_external = X\n        if use_RFE:\n            subset_sklearn, subset_external = self.apply_RFECV(sklearn_model = self.model_sklearn,external_model=self.model_sklearn,X= X,y= y)\n            X_sklearn = X[subset_sklearn]\n            X_external = X[subset_external]\n\n            print(f\"SUBSET {subset_sklearn}\")\n\n        self.feature_columns_sklearn = X_sklearn.columns\n        self.feature_columns_external = X_external.columns\n\n        skf = StratifiedKFold(n_splits=self.n_splits, shuffle=True, random_state=42)\n\n        self.use_resampling = use_resampling\n\n\n        for fold, (train_idx, test_idx) in enumerate(skf.split(X_sklearn, y)):\n\n            print(f\"\\n===== Fold {fold+1}/{self.n_splits} =====\")\n\n            X_train, X_test = X_sklearn.iloc[train_idx], X_sklearn.iloc[test_idx]\n            y_train, y_test = y.iloc[train_idx], y.iloc[test_idx]\n\n            # Tune hyperparameters and train models\n            model_sklearn = self._train_and_predict(\n                X_train, y_train,\n                X_test, y_test,\n                self.feature_columns_sklearn,\n                use_RFE=use_RFE,\n                model=self.model_sklearn,\n                imputation_strategy=imputation_strategy\n            )\n            # Store trained models for later majority vote\n            self.sklearn_group.append(model_sklearn)\n\n\n\n        for fold, (train_idx, test_idx) in enumerate(skf.split(X_external, y)):\n            print(f\"\\n===== Fold {fold+1}/{self.n_splits} =====\")\n\n            X_train, X_test = X_external.iloc[train_idx], X_external.iloc[test_idx]\n            y_train, y_test = y.iloc[train_idx], y.iloc[test_idx]\n\n            # Tune hyperparameters and train models\n            model_external= self._train_and_predict(\n                X_train, y_train,\n                X_test, y_test,\n                self.feature_columns_external,\n                use_RFE=use_RFE,\n                model=self.model_external,\n                imputation_strategy=imputation_strategy\n            )\n\n\n            self.external_group.append(model_external)\n\n        print(\"\\nCross-validation complete.\")\n        self.print_summary()\n\n\n    def _make_pipeline(self, model, imputer=None):\n        imputer = SimpleImputer(strategy=\"mean\") \n        return ImbPipeline([\n            (\"imputer\", imputer),\n            (\"scaler\", StandardScaler()),\n            (\"smoteenn\", SMOTEENN(random_state=42) if self.use_resampling else \"passthrough\"),\n            (\"model\", model)\n        ])\n\n    def _finalize_results(self, model, df, y_test, columns_name):\n        \"\"\"\n            Evaluate a trained model (or pipeline) on test data.\n            Stores accuracy, F1-score, predictions, and performs feature importance & error analysis.\n        \"\"\"\n        # Predictions\n        y_pred = model.predict(df)\n\n        acc = accuracy_score(y_test, y_pred)\n        f1 = f1_score(y_test, y_pred, average='weighted')\n        kappa_score = cohen_kappa_score(y_test, y_pred, weights='quadratic')\n\n        model_name = type(model.named_steps['model']).__name__\n\n        self.results[model_name]['accuracy'].append(acc)\n        self.results[model_name]['f1_score'].append(f1)\n        self.results[model_name]['kappa_score'].append(kappa_score)\n\n        self.results[model_name]['preds'].append(y_pred)\n        self.results[model_name]['true'].append(y_test)\n        self.results[model_name]['estimators'].append(model)\n\n        print(f\"{model_name}: Accuracy={acc:.4f}, F1-weighted={f1:.4f}\")\n        print(f\"Test Set Cohen Kappa (Quadratic): {kappa_score:.4f}\")\n\n        self.compute_feature_importance(model, columns_name)\n        self.analyze_errors(model, df, y_test)\n\n    def analyze_errors(self, model, X_test, y_test):\n        \"\"\"\n            Analyzes predictions: prints metrics, confusion matrix, and error counts per class.\n        \"\"\"\n        print(\"\\nAnalyzing Prediction Errors...\")\n\n        y_pred = model.predict(X_test)\n\n        cm = confusion_matrix(y_test, y_pred)\n        disp = ConfusionMatrixDisplay(cm)\n        disp.plot(cmap='Blues')\n        plt.title(f\"{type(model.named_steps['model']).__name__} - Confusion Matrix\")\n        plt.show()\n\n        class_labels = np.unique(y_test)\n        errors_mask = (y_pred != y_test)\n        print(f\"\\nErrors per class ({type(model.named_steps['model']).__name__}):\")\n        for cls in class_labels:\n            errors = np.sum(errors_mask & (y_test == cls))\n            print(f\"Class {cls}: {errors} errors\")\n        \n        \n    \n    def compute_feature_importance(self, model, feature_columns_used):\n        \"\"\"\n            Computes feature importance for a trained model.\n            Works with pipeline-wrapped models (RandomForest, XGBoost, etc.).\n        \"\"\"\n        print(\"\\nComputing Feature Importance...\")\n\n        final_model = model.named_steps['model']\n\n        feat_imp = pd.DataFrame({\n            'Feature': self.feature_columns_sklearn,\n            'Importance': final_model.feature_importances_\n        }).sort_values(by='Importance', ascending=False)\n\n        feat_imp.plot.bar(x='Feature', y='Importance', figsize=(10, 5), legend=False, title=\"Feature Importance\")\n        plt.title(f'{model.named_steps[\"model\"]} - Feature importance')\n        plt.show()\n        return feat_imp\n\n\n    def print_summary(self):\n        \"\"\"\n        Print cross-validation summary across all folds and all models.\n        Displays mean + std for Accuracy and F1-score.\n        \"\"\"\n        print(\"\\n\" + \"=\"*20 + \" CROSS-VALIDATION SUMMARY \" + \"=\"*20)\n        for model_name, metrics in self.results.items():\n            mean_acc = np.mean(metrics['accuracy'])\n\n            std_acc = np.std(metrics['accuracy'])\n            mean_f1 = np.mean(metrics['f1_score'])\n            std_f1 = np.std(metrics['f1_score'])\n\n            kappa_score_mean = np.mean(metrics['kappa_score'])\n            kappa_score_std = np.std(metrics['kappa_score'])\n\n            print(f\"\\nModel: {model_name}\")\n            print(f\"  Avg. Accuracy: {mean_acc:.4f} ± {std_acc:.4f}\")\n            print(f\"  Avg. F1-score : {mean_f1:.4f} ± {std_f1:.4f}\")\n            print(f\"  Avg. Kappa-Score : {kappa_score_mean:.4f} ± {kappa_score_std:.4f}\")\n\n        print(\"=\"*60)\n\n    def _train_and_predict(self, X_train, y_train,\n                           X_test, y_test,\n                           feature_columns,\n                           use_RFE=True,\n                           model = None,\n                           imputation_strategy='mean'):\n        \"\"\"\n        Train and evaluate both sklearn and external models on a fold.\n        Handles hyperparameter tuning, optional RFECV, and evaluation.\n        Returns the trained models.\n        \"\"\"\n\n        print(\"Tuning and evaluating models...\")\n\n        # ------------------------\n        best_pipe = self._tune_hyperparameters(\n            clone(model),\n            self.param_grid_sklearn,\n            X_train, y_train,\n            imputation_strategy=imputation_strategy\n        )\n\n\n        self._finalize_results(model=best_pipe,\n                                   df=X_test, y_test=y_test,\n                                   columns_name=feature_columns)\n\n        return best_pipe\n\n\n    def _tune_hyperparameters(self, model, params, X_train, y_train,\n                             **kwargs):\n        pipeline = self._make_pipeline(model, kwargs.get(\"imputer\"))\n\n        ## check if params are passed without mode__ since\n        # pipeline is passed\n\n       \n        if params:\n            params = {\n                (k if k.startswith(\"model__\") else f\"model__{k}\"): v\n                for k, v in params.items()\n            }\n            \n        if not params:\n            pipeline.fit(X_train, y_train)\n            return pipeline\n\n        if self.external_search_type == \"random\":\n            best_pipe = RandomizedSearchCV(\n                pipeline, param_distributions=params,\n                n_iter=self.n_iter_external, cv=StratifiedKFold(n_splits=self.hyper_cv, shuffle=True, random_state=42),\n                scoring=\"f1_weighted\", n_jobs=-1, random_state=42\n            )\n        else:\n            best_pipe = GridSearchCV(\n                pipeline, param_grid=params,\n                cv=StratifiedKFold(n_splits=self.hyper_cv, shuffle=True, random_state=42), scoring=\"f1_weighted\", n_jobs=-1\n            )\n\n        best_pipe.fit(X_train, y_train)\n        return best_pipe.best_estimator_\n\n    # --------------------------------------------------------\n\n    def _predict(self,X, method=\"majority\", model_type=\"external\"):\n        \"\"\"\n        Get predictions from the stored models (sklearn or external).\n        Supports majority voting across folds.\n        \"\"\"\n        models = self.external_group if model_type == \"external\" else self.sklearn_group\n        feature_subset = self.feature_columns_external if model_type == \"external\" else self.feature_columns_sklearn\n\n        if not models:\n            raise ValueError(f\"No models found in group '{model_type}'. Did you run start_pipeline()?\")\n\n        print(feature_subset)\n\n        preds = np.array([model.predict(X[feature_subset]) for model in models])\n\n        if method == \"majority\":\n            majority_preds, _ = mode(preds, axis=0, keepdims=False)\n            return majority_preds\n        else:\n            raise ValueError(f\"Unknown method: {method}\")\n\n    def predict(self, X_test, y_test,model_type = \"external\"):\n        \"\"\"\n        Analyzes predictions: prints metrics, confusion matrix, and error counts per class.\n        Works with pipeline-wrapped models.\n        \"\"\"\n        print(\"\\nAnalyzing Prediction Errors...\")\n\n        y_pred = self._predict(X=X_test,model_type=model_type,method=\"majority\")\n\n        cm = confusion_matrix(y_test, y_pred)\n        disp = ConfusionMatrixDisplay(cm)\n        disp.plot(cmap='Blues')\n        plt.title(f\"{model_type} - Confusion Matrix\")\n        plt.show()\n\n        class_labels = np.unique(y_test)\n        errors_mask = (y_pred != y_test)\n        print(f\"\\nErrors per class ({model_type}):\")\n        for cls in class_labels:\n            errors = np.sum(errors_mask & (y_test == cls))\n            print(f\"Class {cls}: {errors} errors\")\n\n        acc = accuracy_score(y_test, y_pred)\n        f1 = f1_score(y_test, y_pred, average='weighted')\n        kappa_score = cohen_kappa_score(y_test, y_pred, weights='quadratic')\n\n        print(f'Accuracy: {acc}')\n        print(f'F1-weighted :{f1}')\n        print(f'Kappa-Score :{kappa_score}')\n\n        return y_pred\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Sono stati testati i seguenti iperparametri:\n\nPer sklearn.RandomForest:\n\n- n_estimatos: numero di alberi da aggregare\n- max_depth: massima profondità dell'albero\n- ccp_alpha: fattore di bilanciamento della complessità dell'albero e della qualità dello split, permette di fare pruning dell'albero\n- criterion: criterio per valutare la bontà dello split\n\nPer XGBoost:\n\n- n_estimatos: numero di alberi da aggregare\n- max_depth: massima profondità dell'albero\n- gamma: fattore di bilanciamento della complessità dell'albero e della qualità dello split, permette di fare pruning dell'albero\n- lambda: termine di L2 regularization. L'aumento di questo valore renderà il modello più conservativo.\n\n\n","metadata":{}},{"cell_type":"code","source":"target = 'sii'\nidentifier = 'id'\n\n# --- Main Script ---\n\ntraining_data = load_clean_training_data()\ntraining_data = training_data.drop(columns=[identifier] + leaky_features)\n\nparam_grid_rf = {\n    \"model__n_estimators\": [100, 200],\n    \"model__max_depth\": [5, 10, None],\n    \"model__ccp_alpha\": [0.0, 0.001, 0.01],\n    \"model__criterion\": ['gini','entropy']\n}\n\n\nxgb_params = {\n       'model__n_estimators': [50, 100, 200],\n        'model__max_depth': [3, 5, 7],\n        \"model__gamma\": [0, 0.1, 0.2],\n        'model__lambda' : [1,5,10,20],\n        'model__eval_metric' : ['rmsle']\n}\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## Test di diversi setting della pipeline: ##","metadata":{}},{"cell_type":"code","source":"lr = RandomForestClassifier(random_state=42)\n\n\nxgb = XGBClassifier(random_state=42, eval_metric='logloss')\n\npipeline_RE = ModelComparisonPipeline(\n    model_sklearn=lr,\n    model_external=xgb,\n    param_grid_sklearn=param_grid_rf,\n    param_grid_external=xgb_params,\n    external_search_type=\"random\",\n    n_iter_external=10\n)\n\npipeline_RE.start_pipeline(df = training_data, use_resampling=True,use_RFE = False)","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"\n\nlr = RandomForestClassifier(random_state=42)\nxgb = XGBClassifier(random_state=42, eval_metric='logloss')\n\npipeline_RFE_RE = ModelComparisonPipeline(\n    model_sklearn=lr,\n    model_external=xgb,\n    param_grid_sklearn=param_grid_rf,\n    param_grid_external=xgb_params,\n    external_search_type=\"random\",\n    n_iter_external=10\n\n)\n\n\npipeline_RFE_RE.start_pipeline(df = training_data, use_resampling=True,use_RFE = True)\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"lr = RandomForestClassifier(random_state=42,class_weight= 'balanced_subsample')\nxgb = XGBClassifier(random_state=42, eval_metric='logloss')\n\npipeline_no_RE = ModelComparisonPipeline(\n    model_sklearn=lr,\n    model_external=xgb,\n    param_grid_sklearn=param_grid_rf,\n    param_grid_external=xgb_params,\n    external_search_type=\"random\",\n    n_iter_external=10\n)\n\npipeline_no_RE.start_pipeline(df = training_data, use_resampling=False,use_RFE = False )","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"lr = RandomForestClassifier(random_state=42,class_weight= 'balanced_subsample')\nxgb = XGBClassifier(random_state=42, eval_metric='logloss')\n\npipeline_no_RE_RFE = ModelComparisonPipeline(\n    model_sklearn=lr,\n    model_external=xgb,\n    param_grid_sklearn=param_grid_rf,\n    param_grid_external=xgb_params,\n    external_search_type=\"random\",\n    n_iter_external=10\n)\n\npipeline_no_RE_RFE.start_pipeline(df = training_data, use_resampling=False,use_RFE = True )\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Test sulle previsioni:","metadata":{}},{"cell_type":"code","source":"test_data = load_clean_test_data()\nX_test = test_data\nY_test = pd.read_csv('/kaggle/input/child-mind-institute-problematic-internet-use/sample_submission.csv')['sii']","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"pred = pipeline_RFE_RE.predict(X_test = X_test, y_test = Y_test,model_type = \"sklearn\")","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"pred2 = pipeline_RE.predict(X_test = X_test, y_test = Y_test,model_type = \"sklearn\")","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"pred3 = pipeline_no_RE.predict(X_test = X_test, y_test = Y_test,model_type = \"sklearn\")","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"pred4 = pipeline_no_RE_RFE.predict(X_test = X_test, y_test = Y_test,model_type = \"sklearn\")","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"pred5 = pipeline_RE.predict(X_test = X_test, y_test = Y_test)\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"pred6 =pipeline_RFE_RE.predict(X_test = X_test, y_test = Y_test,model_type = \"external\")","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"pred7 = pipeline_no_RE.predict(X_test = X_test, y_test = Y_test)","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"pred8 = pipeline_no_RE_RFE.predict(X_test = X_test, y_test = Y_test)","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"results = pd.DataFrame({\n    'id': pd.read_csv('/kaggle/input/child-mind-institute-problematic-internet-use/sample_submission.csv')['id'],\n    'sii': pred7\n})\n\nresults.to_csv('submission.csv', index=False)\nresults\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## EXTRA: Feature Extractions from K-Means, UMAP, Isolation Forest ##\n\nNella seguente parte si cercherà di scovare dei pattern nascosti per migliorare la precisione del nostro modello di classificazione attraverso:\n\n- K-Means : utilizzo di algoritmi non supervisionati per cercare pattern nascosti all'interno dei nostri dati, utilizzeremo il Elbow Method per identificare il numero ottimale di cluster da utilizzare successivamente verrà aggiunto il gruppo trovato come feature all'interno del nostro dataset.\n- Isolation Forest: utilizzo di Isolation Forest per scoprire se la classe meno frequente ( Sii = 3 ) possa essere identificata come anomalia all'interno del nostro dataset cosi da poter utilizzare questa feature per scovare un pattern utile per la classificazione di questa classe.\n- UMAP: permette di generare nuove feature informative senza la creazione manuale di combinazioni di feature e polinomi\n\nContro:\n\nIntroduciamo numerosi iperparametri da tunare, verrà eseguita una random Search per il tuning dei parametri.","metadata":{}},{"cell_type":"markdown","source":"\n![pipeline3.jpg](attachment:84648a66-5271-47df-9114-3c725a96a604.jpg)\n\nPer valutarne la bontà delle feature, ci concentreremo a stimare:\n- bontà di adattamento di Isolation Forest valutandone la precision,recall, F1 sul target considerato.\n- Random Index dei gruppi identificati da K-means\n- Valutazione visiva della proiezione UMAP su 2 dimensioni distinguendole per: classe di sii, anomalie trovate, gruppi dei cluster\n- Stima della correlazione delle nuove feature introdotte sul variabile target","metadata":{},"attachments":{"19598eab-3934-44d6-bdf7-4c3e253a326a.jpg":{"image/jpeg":"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"},"84648a66-5271-47df-9114-3c725a96a604.jpg":{"image/jpeg":"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"}}},{"cell_type":"code","source":"\n\n\nclass FeatureEngineer(BaseEstimator, TransformerMixin):\n    \"\"\"\n    Custom transformer to add features from unsupervised models:\n    - KMeans clusters\n    - IsolationForest anomaly scores\n    - UMAP projections\n    \"\"\"\n    def __init__(self, max_cluster=4, contamination=0.01, n_neighbors=15, min_dist=0.1, random_state=42,target_class = [3],iso_min_f1 = 0.6):\n        self.max_cluster = max_cluster\n        self.contamination = contamination\n        self.n_neighbors = n_neighbors\n        self.min_dist = min_dist\n        self.random_state = random_state\n        self.iso_min_f1 = iso_min_f1\n        self.target_class = target_class\n\n    def fit(self, X, y=None):\n        k_values = range(1,10)\n        errors = []\n\n        for n_c in k_values:\n            kmeans_ = KMeans(n_clusters=n_c, random_state=self.random_state, n_init=10)\n            kmeans_.fit(X)\n\n            errors += [kmeans_.inertia_]\n            \n        # pick the right k_value\n        kneedle = KneeLocator(\n            x=k_values,\n            y=errors,\n            curve=\"convex\",\n            direction=\"decreasing\"\n        )\n\n        optimal_k = kneedle.elbow\n        print(f\"The optimal number of clusters (elbow point) is: {optimal_k}\")\n\n        self.kmeans_ =  KMeans(n_clusters=optimal_k, random_state=self.random_state, n_init=10)\n        self.kmeans_.fit(X)\n\n        self.iso_ = IsolationForest(contamination=self.contamination, random_state=self.random_state)\n        self.iso_.fit(X)\n\n        self.include_iso_feature_ = False\n        \n        if y is not None:\n            y_bin = y.isin(self.target_class).astype(int)\n            iso_preds = (self.iso_.predict(X) == -1).astype(int)\n            prec = precision_score(y_bin, iso_preds, zero_division=0)\n            rec = recall_score(y_bin, iso_preds)\n            f1 = f1_score(y_bin, iso_preds, zero_division=0)\n            print(f\"IsolationForest validation - Precision: {prec:.3f}, Recall: {rec:.3f}, F1: {f1:.3f}\")\n            cm = confusion_matrix(y_bin, iso_preds)\n            print(\"Confusion matrix ISO-Forest (rows=verità, cols=preds):\\n\", cm)\n                \n\n        self.umap_ = umap.UMAP(n_components=2, n_neighbors=self.n_neighbors, min_dist=self.min_dist, random_state=self.random_state)\n        self.umap_.fit(X)\n\n        return self\n\n    def transform(self, X):\n        X_transformed = pd.DataFrame(X).copy()\n\n        X_transformed['_km_cluster'] = self.kmeans_.predict(X)\n\n        iso_preds = self.iso_.predict(X)\n        X_transformed['_is_anomaly'] = (iso_preds == -1).astype(int)\n\n        umap_proj = self.umap_.transform(X)\n        X_transformed['_umap1'] = umap_proj[:, 0]\n        X_transformed['_umap2'] = umap_proj[:, 1]\n\n        \n        return X_transformed.values\n\n    \n    def plot_proj(X,y):\n        # Impostazioni generali\n        plt.figure(figsize=(12, 5))\n        \n        # -------------------------\n        # 1) Plot UMAP con colori delle classi reali\n        # -------------------------\n        plt.subplot(1, 3, 1)\n        if \"_umap1\" in X.columns and \"_umap2\" in X.columns:\n            # Colori basati sulle etichette originali (y_train)\n            scatter = plt.scatter(\n                X[\"_umap1\"], X[\"_umap2\"],\n                c=y.astype(int), cmap=\"tab10\", alpha=0.7\n            )\n            plt.title(\"UMAP projection colored by KMeans clusters\")\n            plt.xlabel(\"_umap1\")\n            plt.ylabel(\"_umap2\")\n            plt.legend(*scatter.legend_elements(), title=\"Clusters\")\n        else:\n            print(\"UMAP projection not found\")\n        \n        # -------------------------\n        # 2) Evidenziare anomalie\n        # -------------------------\n        plt.subplot(1, 3, 2)\n        if \"_umap1\" in X.columns and \"_umap2\" in X.columns:\n            colors = X[\"_is_anomaly\"].map({0: \"grey\", 1: \"red\"})\n            plt .scatter(\n                X[\"_umap1\"], X[\"_umap2\"],\n                c=colors, alpha=0.7\n            )\n            plt.title(\"UMAP projection with anomalies\")\n            plt.xlabel(\"_umap1\")\n            plt.ylabel(\"_umap2\")\n            from matplotlib.lines import Line2D\n            legend_elements = [Line2D([0], [0], marker='o', color='w', label='Normal', markerfacecolor='grey', markersize=8),\n                               Line2D([0], [0], marker='o', color='w', label='Anomaly', markerfacecolor='red', markersize=12)]\n            plt.legend(handles=legend_elements)\n        else:\n            print(\"UMAP projection not found\")\n        \n        # -------------------------\n        # 3) SII\n        # -------------------------\n\n        if y is not None:\n            plt.subplot(1, 3, 3)\n            colors = y.map({0: \"blue\", 1: \"red\",2:\"green\",3:\"orange\"})\n            plt.scatter(\n                X[\"_umap1\"], X[\"_umap2\"],\n                    c=colors, alpha=0.7\n                )\n            plt.title(\"Sii projection\")\n            plt.xlabel(\"_umap1\")\n            plt.ylabel(\"_umap2\")\n            from matplotlib.lines import Line2D\n            legend_elements = [Line2D([0], [0], marker='o', color='w', label='Class:0', markerfacecolor='blue', markersize=8),\n                            Line2D([0], [0], marker='o', color='w', label='Class:1', markerfacecolor='red', markersize=9),\n                              Line2D([0], [0], marker='o', color='w', label='Class:2', markerfacecolor='green', markersize=10),\n                              Line2D([0], [0], marker='o', color='w', label='Class:3', markerfacecolor='orange', markersize=11)]\n            plt.legend(handles=legend_elements)\n                \n        plt.tight_layout()\n        plt.show()","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"training_data = load_clean_training_data()\ntraining_data = training_data.drop(columns=[identifier] + leaky_features)\n\ny_train = training_data['sii']\nX_train = training_data.drop(columns='sii')\n\nfeature_used = X_train.columns.tolist()\n\nfe = FeatureEngineer(random_state=42, target_class=[3], iso_min_f1=0.001)\n\nimputer = SimpleImputer(strategy=\"mean\")\n\nscaler = StandardScaler()\nscaler.fit_transform(X_train)\n\nX_train = pd.DataFrame(imputer.fit_transform(X_train),columns=X_train.columns)\n\nfe.fit(X_train, y_train)\n\nX_transformed = fe.transform(X_train)\n\nX_transformed_df = pd.DataFrame(\n    X_transformed,\n    columns=list(X_train.columns) + ['_km_cluster','_is_anomaly','_umap1','_umap2']\n)\n\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"\n # Impostazioni generali\nplt.figure(figsize=(12, 5))\n        \n# -------------------------\n# 1) Plot UMAP con colori delle classi reali\n# -------------------------\nplt.subplot(2, 3, 1)\nscatter = plt.scatter(\n    X_transformed_df[\"_umap1\"], X_transformed_df[\"_umap2\"],\n    c=y_train.astype(int), cmap=\"tab10\", alpha=0.7\n)\nplt.title(\"UMAP projection colored by KMeans clusters\")\nplt.xlabel(\"_umap1\")\nplt.ylabel(\"_umap2\")\nplt.legend(*scatter.legend_elements(), title=\"Clusters\")\n        \n# -------------------------\n# 2) Evidenziare anomalie\n# -------------------------\nplt.subplot(2, 3, 2)\ncolors = X_transformed_df[\"_is_anomaly\"].map({0: \"grey\", 1: \"red\"})\nplt .scatter(\n    X_transformed_df[\"_umap1\"], X_transformed_df[\"_umap2\"],\n    c=colors, alpha=0.7\n)\nplt.title(\"UMAP projection with anomalies\")\nplt.xlabel(\"_umap1\")\nplt.ylabel(\"_umap2\")\nfrom matplotlib.lines import Line2D\nlegend_elements = [Line2D([0], [0], marker='o', color='w', label='Normal', markerfacecolor='grey', markersize=8),\n                        Line2D([0], [0], marker='o', color='w', label='Anomaly', markerfacecolor='red', markersize=12)]\nplt.legend(handles=legend_elements)\n\n# -------------------------\n# 3) SII\n# -------------------------\n\nplt.subplot(2, 3, 3)\ncolors = y_train.map({0: \"blue\", 1: \"red\",2:\"green\",3:\"orange\"})\nplt.scatter(\n            X_transformed_df[\"_umap1\"], X_transformed_df[\"_umap2\"],\n                    c=colors, alpha=0.7\n                )\nplt.title(\"Sii projection\")\nplt.xlabel(\"_umap1\")\nplt.ylabel(\"_umap2\")\nfrom matplotlib.lines import Line2D\nlegend_elements = [Line2D([0], [0], marker='o', color='w', label='Class:0', markerfacecolor='blue', markersize=8),\n                    Line2D([0], [0], marker='o', color='w', label='Class:1', markerfacecolor='red', markersize=9),\n                      Line2D([0], [0], marker='o', color='w', label='Class:2', markerfacecolor='green', markersize=10),\n                    Line2D([0], [0], marker='o', color='w', label='Class:3', markerfacecolor='orange', markersize=15)]\nplt.legend(handles=legend_elements)\n                \nplt.tight_layout()\nplt.show()\n\n\n# -------------------------\n# 4) SII\n# -------------------------\n\nplt.subplot(2, 1, 1)\ncolors = y_train.map({0: \"grey\", 1: \"grey\",2:\"green\",3:\"orange\"})\nplt.scatter(\n            X_transformed_df[\"_umap1\"], X_transformed_df[\"_umap2\"],\n                    c=colors, alpha=0.7\n                )\nplt.title(\"Sii projection\")\nplt.xlabel(\"_umap1\")\nplt.ylabel(\"_umap2\")\nfrom matplotlib.lines import Line2D\nlegend_elements = [Line2D([0], [0], marker='o', color='w', label='Class:0', markerfacecolor='blue', markersize=8),\n                    Line2D([0], [0], marker='o', color='w', label='Class:1', markerfacecolor='red', markersize=9),\n                      Line2D([0], [0], marker='o', color='w', label='Class:2', markerfacecolor='green', markersize=10),\n                    Line2D([0], [0], marker='o', color='w', label='Class:3', markerfacecolor='orange', markersize=15)]\nplt.legend(handles=legend_elements)\n                \nplt.tight_layout()\nplt.show()\n","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Possiamo osservare che nella proiezione UMAP a due dimensioni le classi sono estremamente mischiate tra di loro, rendendo quindi la ricerca di qualsiasi pattern nascosto complicata.","metadata":{}},{"cell_type":"code","source":"from sklearn.metrics import rand_score\n\n# cluster predetti\nclusters = X_transformed_df['_km_cluster']\n\n# quanto i cluster coincidono con le classi reali\n\nari = rand_score(y_train, clusters)\nprint(f\"Rand Index (cluster vs class): {ari:.3f}\")","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"corr = X_transformed_df[['_km_cluster','_umap2','_umap1','_is_anomaly']].copy()\ncorr['sii'] = y_train\n\nsns.heatmap(corr.corr())","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"Le feature non sono significative per la predizione della classe di Sii, rendendole dunque non utili.","metadata":{}},{"cell_type":"code","source":"results = pd.DataFrame({\n    'id': pd.read_csv('/kaggle/input/child-mind-institute-problematic-internet-use/sample_submission.csv')['id'],\n    'sii': pred7\n})\n\nresults.to_csv('submission.csv', index=False)\nresults","metadata":{"trusted":true},"outputs":[],"execution_count":null}]}