{"cells":[{"metadata":{"trusted":true,"_kg_hide-input":true,"_kg_hide-output":true},"cell_type":"code","source":"from ast import literal_eval\nfrom collections import defaultdict \nfrom itertools import groupby\nfrom operator import itemgetter \n\nfrom nltk.corpus import stopwords\nfrom nltk.stem import SnowballStemmer\n\nfrom scipy.interpolate import UnivariateSpline\n\nfrom sklearn.calibration import CalibratedClassifierCV\nfrom sklearn.feature_extraction.text import TfidfVectorizer\nfrom sklearn.metrics import classification_report, log_loss\nfrom sklearn.model_selection import KFold, train_test_split\nfrom sklearn.pipeline import FeatureUnion, make_pipeline\nfrom sklearn.svm import LinearSVC\n\nimport collections\nimport matplotlib.pyplot as plt\nimport numpy as np\nimport os\nimport pandas as pd\nimport re\nimport seaborn as sns\nimport statsmodels.api as sm\nimport xgboost as xgb\nsns.set()\n\nseed = 42\n\nfrom sklearn.metrics import (\n average_precision_score, confusion_matrix, ConfusionMatrixDisplay, plot_confusion_matrix,\n plot_precision_recall_curve, plot_roc_curve, precision_recall_curve, roc_auc_score, roc_curve\n)\n\ndef plot_confusion_matrix_from_proba(estimator, X, y_true, threshold, labels=None,\n sample_weight=None, normalize=None,\n display_labels=None, include_values=True,\n xticks_rotation='horizontal',\n values_format=None,\n cmap='viridis', ax=None):\n\n y_pred = estimator.predict_proba(X)[:, 1] > threshold\n cm = confusion_matrix(y_true, y_pred, sample_weight=sample_weight,\n labels=labels, normalize=normalize)\n\n if display_labels is None:\n if labels is None:\n display_labels = estimator.classes_\n else:\n display_labels = labels\n\n disp = ConfusionMatrixDisplay(confusion_matrix=cm,\n display_labels=display_labels)\n return disp.plot(include_values=include_values,\n cmap=cmap, ax=ax, xticks_rotation=xticks_rotation,\n values_format=values_format)\n\n\ndef plot_classification_results(clf, X_test, y_test, y_proba=None, threshold=None, nb_disp_thresh=10, harmonic_opt=True):\n fig, ax = plt.subplots(figsize=(15, 4), ncols=3)\n\n if threshold is None:\n titles_options = [\n (\"Confusion matrix\", None, \".0f\"), # , without normalization\n # (\"Normalized confusion matrix\", 'true', \".4f\")\n ]\n else:\n titles_options = [\n (\"Confusion matrix with threshold {:.4f}\".format(threshold), None, \".0f\"), # ,\\n without normalization\n # (\"Normalized confusion matrix\\n with threshold {:.4f}\".format(threshold), 'true', \".4f\")\n ]\n\n for ind, (title, normalize, values_format) in enumerate(titles_options):\n if threshold is None:\n plot_confusion_matrix(\n clf, X_test, y_test, display_labels=['0', '1'], cmap=plt.cm.Blues, normalize=normalize,\n values_format=values_format, ax=ax[ind]\n )\n else:\n plot_confusion_matrix_from_proba(\n clf, X_test, y_test, threshold=threshold, display_labels=['0', '1'], cmap=plt.cm.Blues,\n normalize=normalize, values_format=values_format, ax=ax[ind]\n )\n ax[ind].set_title(title)\n ax[ind].grid(False)\n\n if y_proba is None:\n y_proba = clf.predict_proba(X_test)[:, 1]\n\n plot_roc_curve(clf, X_test, y_test, ax=ax[1], label=\"ROC curve\")\n \n fpr, tpr, thresholds = roc_curve(y_test, y_proba)\n step = len(fpr) // (nb_disp_thresh - 1) if len(fpr) // (nb_disp_thresh - 1) >= 2 else 1\n\n ax[1].plot(fpr[1:-1:step], tpr[1:-1:step], 'o', label=\"Thresholds\")\n for x, y, txt in zip(fpr[1:-1:step], tpr[1:-1:step], thresholds[1:-1:step]):\n ax[1].annotate(np.round(txt, 2), (x - 0.08, y + 0.02))\n \n if harmonic_opt:\n argmax = np.argmax(2 * tpr * (1 - fpr) / (tpr + (1 - fpr) + 1e-6))\n else:\n argmax = np.argmax(tpr + (1 - fpr))\n ax[1].plot(fpr[argmax], tpr[argmax], 'o', label=\"Best threshold\", color=\"green\")\n ax[1].annotate(np.round(thresholds[argmax], 4), (fpr[argmax] + 0.01, tpr[argmax] - 0.06), color=\"green\")\n\n ax[1].plot([0, 1], [0, 1], linestyle='--', color='r', label='Chance')\n ax[1].set_title(\"ROC curve (AUC = {:.4f})\".format(roc_auc_score(y_test, y_proba)))\n ax[1].legend()\n\n plot_precision_recall_curve(clf, X_test, y_test, ax=ax[2], label=\"PR curve\")\n pr, rc, thresholds = precision_recall_curve(y_test, clf.predict_proba(X_test)[:, 1])\n step = len(rc) // (nb_disp_thresh - 1) if len(rc) // (nb_disp_thresh - 1) >= 2 else 1\n \n ax[2].plot(\n rc[::step], pr[::step], 'o', label=\"Thresholds\"\n )\n for x, y, txt in zip(rc[1:-1:step], pr[1:-1:step], thresholds[1:-1:step]):\n ax[2].annotate(np.round(txt, 2), (x, y + 0.02))\n\n if harmonic_opt:\n argmax = np.argmax(2 * pr * rc / (pr + rc + 1e-6))\n else:\n argmax = np.argmax(pr + rc)\n ax[2].plot(rc[argmax], pr[argmax], 'o', label=\"Best threshold\", color=\"green\")\n ax[2].annotate(np.round(thresholds[argmax], 4), (rc[argmax] - 0.18, pr[argmax] - 0.03), color=\"green\")\n\n ax[2].set_title(\"PR curve (AP = {:.4f})\".format(average_precision_score(y_test, y_proba)))\n ax[2].legend()\n ax[2].set_xlim(-0.05, 1.05)\n ax[2].set_ylim(-0.05, 1.05)\n\n plt.show()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Madness at Home and on the Court - Part 2\n\n*Authors: [Emilien Etchevers](https://www.kaggle.com/emimis), [Kieran Janin](https://www.kaggle.com/kieranjanin), [Michael Karpe](https://www.kaggle.com/mika30), [Remi Le Thai](https://www.kaggle.com/remilethai), [Haley Wohlever](https://www.kaggle.com/haleywohlever)*\n\n# Contents\n\n*In the [first notebook](https://www.kaggle.com/emimis/madness-at-home-and-on-the-court-part-1)*\n\n- [Introduction](#Introduction)
\n- [NCAA March Madness Data Analysis](#NCAAMarchMadnessData)
\n * [Seeding and Entertainment](#SeedingandEntertainment)
\n * [Madness through Unpredictability](#MadnessthroughUnpredictability)
\n * [Entertainment due to Closeness of Games](#EntertainmentduetoClosenessofGames)\n \n*In this notebook*\n\n- [Tweets on NCAA Data Analysis](#TweetsonNCAAData)
\n * [Temporal Evolution of Engagement](#TemporalEvolutionofEngagement)
\n * [Team Mentions Count in Tweets](#TeamMentionsCountinTweets)
\n * [Sentiment Analysis for Tweets on NCAA](#SentimentAnalysisforTweetsonNCAA)\n- [Conclusion](#Conclusion)\n\n\n# 3. Tweets on NCAA Data\n\nAfter thoroughly exploring the NCAA data, we decided to extend our analysis by studying relevant tweets published during the 2015 to 2019 NCAA March Madness tournaments. We built a database of 1.6 million tweets containing the word *NCAA*, using the [GetOldTweets3](https://github.com/Mottl/GetOldTweets3) library. The dataset we built with this library is available [here](https://www.kaggle.com/mika30/ncaa-tweets-20152019) (as well as a processed version [here](https://www.kaggle.com/mika30/ncaa-tweets-processed) to avoid memory issues).\n\nThe purpose of this second part of our analysis is to identify trends in tweets published during and related to the competition. As discussed in Part 1, we want to distinguish an *objective* madness, explainable by match and player data, from a *subjective* madness that can be observed in the public sentiment surrounding the games.\n\n\n## 3.1. Temporal Evolution of Engagement\n\nWe first study *engagement* on Twitter. *Engagement* on Twitter is defined by three types of actions: retweets, replies, and favorites (or likes). We plot the evolution of the number of each of these actions over the 21 days of the competition for each season, as well as the evolution of the number of tweets.\n\nWhile we can see that the number of tweets including the word *NCAA* tends to slightly decrease over the years (or at least remains constant), we can observe an increase in the number of retweets and replies over the years. We also note an even higher increase in the number of favorites, this latest observation showing that people tend to express themselves more and more using the favorites feature on Twitter."},{"metadata":{"trusted":true,"_kg_hide-input":true},"cell_type":"code","source":"processing = False\n\nif processing:\n tweets = pd.read_csv('/kaggle/input/ncaa-tweets-20152019/ncaa-tweets-2015-2019.csv', parse_dates=[\"date\"])\n tweets[\"date\"] = tweets[\"date\"] - pd.Timedelta(hours=4) # UTC to ET\n tweets = tweets[tweets[\"date\"].apply(\n lambda x: str(x)[:10] not in [\"2015-03-16\", \"2016-03-14\", \"2017-03-13\", \"2018-03-12\", \"2019-03-18\"]\n )]\nelse:\n tweets = pd.read_csv(\n '/kaggle/input/ncaa-tweets-processed/tweets-processed.csv', parse_dates=[\"date\"],\n converters={\"TeamIDMatch\": literal_eval, \"GameandTeamMatch\": literal_eval}\n )\n\ndisplay(tweets.sample(5))\n\ngbcount = tweets.groupby([tweets['date'].dt.date]).count()\ngbsumday = tweets.groupby([tweets['date'].dt.date])[\"replies\", \"retweets\", \"favorites\"].sum()\n\ncolors = ['r', 'g', 'b', 'purple', 'orange']\n\nbegin_year, end_year = 2015, 2019\nbegin_day, end_day = 1, 21\n# begin_day, end_day = 134, 154\nnb_days = end_day - begin_day + 1\n\nfig, ax = plt.subplots(figsize=(15, 8), nrows=2, ncols=2)\nfor axis, engagement in enumerate([\"tweets\", \"replies\", \"retweets\", \"favorites\"]):\n for ind, year in enumerate(range(begin_year, end_year + 1)):\n if engagement == \"tweets\":\n ax[axis // 2, axis % 2].plot(\n range(1, nb_days + 1), gbcount[\"date\"][nb_days * ind:nb_days * (ind + 1)],c=colors[ind], label=year\n )\n else:\n ax[axis // 2, axis % 2].plot(\n range(1, nb_days + 1), gbsumday[engagement][nb_days * ind:nb_days * (ind + 1)], c=colors[ind], label=year\n )\n ax[axis // 2, axis % 2].set_xticks(range(1, nb_days + 1))\n ax[axis // 2, axis % 2].set_xlabel(\"Day of the competition\")\n ax[axis // 2, axis % 2].set_ylabel(\"Number of \" + engagement)\n # ax[axis // 2, axis % 2].set_title(\"Number of \" + engagement)\n ax[axis // 2, axis % 2].legend(loc=\"upper center\")\n\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"If we consider the number of favorites as a key indicator of the madness generated by the competition, we can remark that the major spikes in favorites reflect the March Madness schedule. Spikes at the beginning of the competition correspond to *First Round* and *Second Round*, spikes in the middle to *Sweet Sixteen* and *Elite Eight*, and in the end to the *Final Four*.\n\nThe lack of spikes when there are no games shows that madness is generated by the games themselves. Moreover, the much higher number of favorites compared to the number of tweets, replies, and retweets indicates that NCAA March Madness creates positive reactions. If there were more negative comments in reaction to games, it is likely people would feel the need to reply to either also express their frustration or to argue back against someone rather than simply favorite the initial tweet.\n\n\n## 3.2. Team Mentions Count in Tweets\n\nIn this section, we aim to refine our study by focusing on the number of team mentions per team and per day. Using tables of team spellings, player names, and coach names, we tried to find tweets mentioning specific teams or people. We present these results in the form of heatmaps, the rows corresponding to teams and the columns to the day of the competition."},{"metadata":{"trusted":true,"_kg_hide-input":true},"cell_type":"code","source":"# stop_words = stopwords.words(\"english\")\n# stemmer = SnowballStemmer('english')\n\ndef clean_tweet(tweet):\n tweet = tweet.lower()\n tweet = re.sub(r\"http\\S+\", \"\", tweet) # remove hyperlinks\n tweet = re.sub(r\"@\\S+\", \"\", tweet) # remove user names after @\n tweet = re.sub(r\"&\\w+;\", \"\", tweet) # remove first words after & (HTML chars)\n tweet = re.sub(r\"\\n\", \" \", tweet) # remove newlines\n # tweet = re.sub(r\"[^\\w\\s]\", \"\", tweet) # remove special characters\n tweet = re.sub(r\"[^A-Za-z0-9?,.;/:!\\)\\(\\-\\s]+\", \" \", tweet) # keep letters, numbers, spaces and some punctuation\n tweet = re.sub(r\"([\\w\\s-]+|[^\\w\\s-]+)\\s*\", r\"\\1 \", tweet) # separate words from punctuation except for -\n tweet = re.sub(r\" +\", \" \", tweet) # remove multiple whitespaces\n tweet = re.sub(r\"^\\s+|\\s+$\", \"\", tweet) # remove leading and trailing spaces\n # tweet = \" \".join([word for word in tweet.split() if word not in stop_words]) # remove stop words\n # tweet = \" \".join([stemmer.stem(word) for word in tweet.split()]) # stemming\n return tweet\n\n\nif processing:\n tweets[\"processed_text\"] = tweets[\"text\"].apply(clean_tweet)\n tweets = tweets[tweets[\"processed_text\"] != \"\"].drop_duplicates([\"processed_text\"]).reset_index(drop=True)\n tweets[\"Season\"] = pd.to_datetime(tweets[\"date\"]).dt.year\n\n\nPATH_TO_MEN_DATA = \"../input/march-madness-analytics-2020/2020DataFiles/2020DataFiles/2020-Mens-Data\"\nPATH_TO_STAGE2_DATA = \"../input/march-madness-analytics-2020/MDataFiles_Stage2\"\n\nteam_spellings = pd.read_csv(PATH_TO_STAGE2_DATA + \"/MTeamSpellings.csv\", encoding='latin1')\nteam_players = pd.read_csv(PATH_TO_MEN_DATA + \"/MPlayers.csv\")\nteam_coaches = pd.read_csv(PATH_TO_STAGE2_DATA + \"/MTeamCoaches.csv\")\n\nteam_id_to_spellings = team_spellings.groupby('TeamID')['TeamNameSpelling'].apply(lambda x: list(set(x))).to_dict()\nteam_id_to_players = team_players.groupby('TeamID')['LastName'].apply(lambda x: list(set(x))).to_dict()\nteam_id_to_coaches = team_coaches.groupby('TeamID')['CoachName'].apply(lambda x: list(set(x))).to_dict()\n\nteam_id_to_tweet_mentions = {\n key: sorted(list(\n possible_name.lower() for possible_name in set(\n value + team_id_to_players.get(key, []) + team_id_to_coaches.get(key, [])\n ) if len(possible_name) >= 4\n )) for key, value in team_id_to_spellings.items()\n}\n\nMEvents = pd.concat([\n pd.read_csv(PATH_TO_MEN_DATA + \"/MEvents{}.csv\".format(year)) for year in range(begin_year, end_year + 1)\n])\nMSeasons = pd.read_csv(PATH_TO_STAGE2_DATA + \"/MSeasons.csv\")\nMNCAATourneySeeds = pd.read_csv(PATH_TO_STAGE2_DATA + \"/MNCAATourneySeeds.csv\")\nMNCAATourneyCompactResults = pd.read_csv(PATH_TO_STAGE2_DATA + \"/MNCAATourneyCompactResults.csv\")\n\n# Dictionary with all teams from 2015 to 2019 \nyear_to_team_ids = {\n year: MNCAATourneySeeds[MNCAATourneySeeds[\"Season\"] == year][\"TeamID\"].tolist()\n for year in range(begin_year, end_year + 1)\n} # NCAATeamsDict\n\n# Dictionary with the starting date of a given season\nyear_to_begin_day = {\n key: pd.to_datetime(value) for key, value in {\n 2015: \"2015-03-17\", 2016: \"2016-03-15\", 2017: \"2017-03-14\", 2018: \"2018-03-13\", 2019: \"2019-03-19\"\n }.items()\n}\n\n# Dictionary with the individual Tourney games with day number, winning team ID and losing team ID\nyear_to_day_to_game_ids = {\n year : [\n (DayNum,WTeamID,LTeamID) \n for DayNum,WTeamID,LTeamID in zip(\n MNCAATourneyCompactResults[MNCAATourneyCompactResults[\"Season\"] == year][\"DayNum\"] - 133,\n MNCAATourneyCompactResults[MNCAATourneyCompactResults[\"Season\"] == year][\"WTeamID\"],\n MNCAATourneyCompactResults[MNCAATourneyCompactResults[\"Season\"] == year][\"LTeamID\"]\n )\n ] for year in range(begin_year, end_year + 1)\n}\n\n# Create embedded dictionnary for each tournement day\nfor year in range(begin_year, end_year + 1):\n year_to_day_to_game_ids[year] = dict(\n (DayNum, [team for Teams in itr for team in Teams[1:]])\n for DayNum, itr in groupby(year_to_day_to_game_ids[year], itemgetter(0))\n ) \n\n\nif processing:\n # Setting up DayNum column \n tweets[\"DayNum\"] = tweets[\"date\"].apply(lambda x: x - year_to_begin_day[int(x.year)]).dt.days + 1\n\n for year in range(begin_year, end_year + 1):\n tweets.loc[tweets[\"Season\"] == year, \"TeamIDMatch\"] = tweets.loc[tweets[\"Season\"] == year, \"text\"].apply(\n lambda tweet: [\n key for key in year_to_team_ids[year]\n if any(keyword in tweet for keyword in team_id_to_tweet_mentions[key])\n ]\n )\n tweets.loc[tweets[\"Season\"] == year, \"GameandTeamMatch\"] = tweets.loc[\n tweets[\"Season\"] == year, [\"TeamIDMatch\", \"Season\", \"DayNum\"]\n ].apply(\n lambda values: list(set(values[0]) & set(year_to_day_to_game_ids[values[1]].get(values[2], []))), axis=1\n )\n\n tweets.to_csv(\"tweets-processed.csv\", index=False)\n\n\ndef plot_team_day_heatmap(year_to_matrix, title):\n fig, ax = plt.subplots(figsize=(14, 15))\n sns.heatmap(\n year_to_matrix[end_year], cmap=plt.cm.Blues, ax=ax, xticklabels=range(1, nb_days + 1),\n yticklabels=[team_id_to_spellings[x][0] for x in year_to_team_ids[end_year]], # cbar=False\n )\n ax.set_title(\"{} in Year {}\".format(title, end_year))\n ax.set_xlabel(\"Day of competition (1st to 21st)\")\n ax.set_ylabel(\"Name of the team\")\n plt.show()\n\n nb_rows, nb_cols = 2, 3\n fig, ax = plt.subplots(figsize=(15, 10), nrows=nb_rows, ncols=nb_cols)\n for ind, year in enumerate(range(begin_year, end_year + 1)):\n sns.heatmap(\n year_to_matrix[year], cmap=plt.cm.Blues, ax=ax[ind // nb_cols, ind % nb_cols],\n xticklabels=False, yticklabels=False, # cbar=False\n )\n ax[ind // nb_cols, ind % nb_cols].set_title(\"{} in Year {}\".format(title, year))\n ax[ind // nb_cols, ind % nb_cols].set_xlabel(\"Day of competition (1st to 21st)\")\n ax[ind // nb_cols, ind % nb_cols].set_ylabel(\"ID of the team\")\n plt.show()\n\n\nyear_to_team_mentions_count = {\n year: np.array([[\n tweets[\n (tweets[\"Season\"] == year) & (tweets[\"DayNum\"] == day)\n ][\"TeamIDMatch\"].apply(lambda x: team_id in x).sum()\n for day in range(begin_day, end_day + 1)\n ] for team_id in year_to_team_ids[year]\n ]) for year in range(begin_year, end_year + 1)\n}\n\nplot_team_day_heatmap(year_to_team_mentions_count, \"Number of mentions\")","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Again, we note for all years that the darkest columns, i.e. the ones with the highest number of mentions in a given day, correspond to the main stages of the tournament. When looking at the rows, we can observe that some teams are mentioned more than others in our set of tweets.\n\nFor example, on the heatmap corresponding to the year 2019, we observe that the team that generated the highest number of mentions on a given day is *Northern Kentucky* on Day 4. When looking at the 2019 schedule as well as press releases, we can confirm that *Northern Kentucky* was playing on 03/22/19. They lost a game against *Texas Tech*, who is known for their solid defense:\n\n| | |\n|-------------------------|-------------------------|\n| ![north-kentucky-1.png](attachment:north-kentucky-1.png) | ![north-kentucky-press.png](attachment:north-kentucky-press.png) |\n\n
\n\nThis new indicator of team mentions on a given day shows that some of the teams likely have more supporters active on Twitter! The number of active supporters on Twitter may then be an indicator for the madness generated by a team.\n\n\n## 3.3. Sentiment Analysis for Tweets on NCAA\n\nIn this final section of our madness exploration, we applied sentiment analysis to the NCAA Tweets, training a classifier on the [Sentiment140](https://www.kaggle.com/kazanova/sentiment140) dataset (which also contains around 1.6 million tweets) and then predicting sentiments on NCAA tweets.\n\nWe display a sample of lines of Sentiment140 dataset, as well as the performance of our classifier on a validation set (with an accuracy around 82.5% and an area under the ROC curve around 0.906) taken from the Sentiment140 dataset. We also display the probability distribution of positive and negative sentiments for the validation set and NCAA tweets.","attachments":{"north-kentucky-1.png":{"image/png":"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"},"north-kentucky-press.png":{"image/png":"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"}}},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true,"_kg_hide-input":true},"cell_type":"code","source":"if processing:\n sentiment140 = pd.read_csv(\n '/kaggle/input/sentiment140/training.1600000.processed.noemoticon.csv', header=None, parse_dates=[2],\n names=[\"sentiment\", \"id\", \"date\", \"flag\", \"username\", \"text\"], encoding=\"ISO-8859-1\"\n )\n sentiment140[\"processed_text\"] = sentiment140[\"text\"].apply(clean_tweet)\n sentiment140 = sentiment140[sentiment140[\"processed_text\"] != \"\"].drop_duplicates([\"processed_text\"]).reset_index(drop=True)\n sentiment140[\"sentiment\"] = sentiment140[\"sentiment\"].apply(lambda x: x // 4)\n sentiment140.to_csv(\"sentiment140-processed.csv\", encoding=\"utf-8\", index=False)\nelse:\n sentiment140 = pd.read_csv('/kaggle/input/ncaa-tweets-processed/sentiment140-processed.csv')\n\ndisplay(sentiment140.sample(5))\n\nX_train, X_val, y_train, y_val = train_test_split(\n sentiment140[\"processed_text\"].apply(str), sentiment140[\"sentiment\"], random_state=seed, test_size=0.2\n)\nX_test = tweets[\"processed_text\"].apply(str)\n\npipe = make_pipeline(\n FeatureUnion([\n ('words', TfidfVectorizer(ngram_range=(1, 3), analyzer='word')),\n ('chars', TfidfVectorizer(ngram_range=(1, 3), analyzer='char')),\n ]),\n CalibratedClassifierCV(LinearSVC(C=1.0, random_state=seed), cv=5),\n)\npipe.fit(X_train, y_train)\n\ny_val_proba = pipe.predict_proba(X_val)[:, 1]\nthreshold = 0.4874\ny_val_pred = y_val_proba > threshold\n\nplot_classification_results(pipe, X_val, y_val, y_proba=y_val_proba, threshold=threshold)\nplt.show()\n\nplt.figure(figsize=(15, 4))\nplt.hist(y_val_proba, bins=100, density=True, label=\"Sentiment140 Tweets\")\n\ny_proba_1 = pipe.predict_proba(X_test[:len(X_test) // 2])[:, 1]\ny_proba_2 = pipe.predict_proba(X_test[len(X_test) // 2:])[:, 1]\ny_proba = np.concatenate([y_proba_1, y_proba_2])\ny_pred = (y_proba > threshold) * 1\n\nplt.hist(y_proba, bins=100, density=True, label=\"NCAA Tweets\", alpha=0.5)\nplt.title(\"Probability density of positive sentiments\")\nplt.xlabel(\"Probability\")\nplt.ylabel(\"Density\")\nplt.legend()\nplt.show()\n\ntweets[\"proba\"] = y_proba\ntweets[\"prediction\"] = y_pred\ntweets.to_csv(\"tweets-sentiments.csv\", index=False)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"While the classifier is able to make a clear distinction between positive and negative tweets on the Sentiment140 dataset, predictions on our NCAA tweets show a more uniform distribution that is skewed left. The fact that the probability distribution on NCAA tweets tends to be more uniform is due to the neutrality of a sizeable proportion of NCAA tweets. However, the skewed-left-characteristic shows that NCAA generates more positive sentiments than negative sentiments overall. This shows that what we call *madness* in the NCAA competition is more related to positive than negative sentiments.\n\nWith the predictions from our classifier, we display below heatmaps corresponding the average sentiment (between 0 and 1) for a given team on a given day. Here, 0 is taken as a negative sentiment, 1 is taken as a positive sentiment, and 0.5 is taken as the threshold between a positive and negative sentiment."},{"metadata":{"trusted":true,"_kg_hide-input":true},"cell_type":"code","source":"year_to_average_sentiment = {\n year: np.array([[\n tweets[\n (tweets[\"Season\"] == year) & (tweets[\"DayNum\"] == day) &\n (tweets[\"TeamIDMatch\"].apply(lambda x: team_id in x))\n ].proba.mean()\n for day in range(begin_day, end_day + 1)\n ] for team_id in year_to_team_ids[year]\n ]) for year in range(begin_year, end_year + 1)\n}\n\nplot_team_day_heatmap(year_to_average_sentiment, \"Average sentiment\")","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Although we do not observe clear distinctions between teams for a positive or a negative sentiment, we see that the average sentiment for a given team on a given day is over 0.5. meaning that overall positive sentiments are more prevalent than negative ones. This follows our previous observation that NCAA March Madness creates positive reactions in its fans.\n\n\n# 4. Conclusion\n\nOur exploration of the NCAA data in Part 1 revealed that there does indeed exist a link between watching higher seeded teams compete and the objective entertainment of those games. Filled with more point-effective shots, a greater degree of player collaboration, and higher scores, we are able to conclude that the madness of a game is correlated to the ranks of the involved teams.\n\nSimilarly, we were able to see trends related to the madness of a match and the closeness of the score throughout. While we feel that the madness cannot truly be predicted from these metrics alone, it in many ways confirms our initial suspicions that much of the delightful madness of this tournament is derived from its inability to be predicted.\n\nThe sentiment analysis we have completed here in Part 2 is telling of the enjoyment of fans. In some ways it may seem that our analysis of Twitter data was unable to uncover any specific and telling trends related to the public’s response to March Madness throughout the season. We see that their social media activity is centered on the main parts of the competition and that the most buzz is generated right at the start when the most teams are involved, and the highest number of games occur. While all of this is relatively intuitive, what is somewhat surprising and worthy of further evaluation is that the overall reaction to these games is positive.\n\nTwitter has in many ways built up a reputation for being a place that people can express short, inflammatory remarks which may be expected when upsets inevitably occur throughout the tournament. However, this is not the trend we saw. As indicated by the trends in positive sentiments expressed on Twitter, even when upsets occur and favorites lose, by and large the some 40 million people following the competition still love the game and we are sure that, after the cancellation of this season, there will be even more people looking forward to watching the madness play out in the 2021 games."}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}