","metadata":{"papermill":{"duration":0.007425,"end_time":"2025-06-11T13:37:19.521105","exception":false,"start_time":"2025-06-11T13:37:19.513680","status":"completed"},"tags":[]}},{"cell_type":"markdown","source":"##
03 :🔬🧐 Data Inspection\n\n
\n
\n Findings \n
\n └ 1 : describe later
\n\n \n Insights \n
\n └ 1 : describe later
\n ","metadata":{"papermill":{"duration":0.007398,"end_time":"2025-06-11T13:37:19.536065","exception":false,"start_time":"2025-06-11T13:37:19.528667","status":"completed"},"tags":[]}},{"cell_type":"code","source":"# Check if each column has a unique value of 0 and 1, and assign 1 or 0\ndef check_if_binary(column):\n unique_values = column.unique()\n if len(unique_values) == 2 and set(unique_values) == {0, 1}:\n return 1\n else:\n return 0","metadata":{"papermill":{"duration":0.013397,"end_time":"2025-06-11T13:37:19.556915","exception":false,"start_time":"2025-06-11T13:37:19.543518","status":"completed"},"tags":[],"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T13:43:48.091562Z","iopub.execute_input":"2025-07-09T13:43:48.092039Z","iopub.status.idle":"2025-07-09T13:43:48.096097Z","shell.execute_reply.started":"2025-07-09T13:43:48.092018Z","shell.execute_reply":"2025-07-09T13:43:48.095409Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"def check_if_outliers(df):\n\n outliers_rate_dict = {}\n\n for column_name in df.columns:\n column = df[column_name]\n\n # Skip if not a numeric column\n if not pd.api.types.is_numeric_dtype(column):\n print(f\"'{column_name}' is skipped because it is not a numeric type.\")\n continue\n\n # Calculate the mean and standard deviation of the data\n mean = column.mean()\n std = column.std()\n\n # Set outlier threshold\n threshold = 2 # Adjust this value to change the outlier criteria\n\n # Set conditions for detecting outliers\n lower_bound = mean - threshold * std\n upper_bound = mean + threshold * std\n\n # Detect outliers\n outliers = (column < lower_bound) | (column > upper_bound)\n\n # Calculate the percentage of outliers\n outliers_rate = outliers.sum() / len(column)\n\n # Save results to dictionary\n outliers_rate_dict[column_name] = outliers_rate\n\n return outliers_rate_dict","metadata":{"papermill":{"duration":0.014551,"end_time":"2025-06-11T13:37:19.579031","exception":false,"start_time":"2025-06-11T13:37:19.564480","status":"completed"},"tags":[],"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T13:43:48.252105Z","iopub.execute_input":"2025-07-09T13:43:48.252601Z","iopub.status.idle":"2025-07-09T13:43:48.257356Z","shell.execute_reply.started":"2025-07-09T13:43:48.252578Z","shell.execute_reply":"2025-07-09T13:43:48.256777Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"def data_inspection(df):\n \"\"\"A function that generates data inspection information for a data frame\"\"\"\n\n # process_one:Create a DataFrame with basic statistics\n data_inspection = pd.DataFrame({\n 'column_name' : df.columns,\n 'data_type' : df.dtypes,\n 'cnt_rows' : len(df),\n 'cnt_unique_rows' : df.nunique(),\n 'cnt_duplicated_rows' : len(df) - df.nunique(),\n 'cnt_non_null_rows' : df.count().values,\n 'cnt_null_rows' : df.isnull().sum(),\n 'rate_null_rows' : (df.isnull().sum() / len(df)),\n # 'date min': df['t_dat'].min(),\n # 'date max': df['t_dat'].max()\n })\n\n # process_two:Calculate descriptive statistics, modes, and percentages of modes for numerical data:\n\n # The below code is no longer needed as 'custom_description' provides the stats\n # description = df.describe(include=np.number).T.reset_index().rename(columns={'index': 'column_name'})\n # median = df.median(numeric_only=True).reset_index().rename(columns={'index': 'column_name', 0: 'median'})\n\n # Calculate the most frequent value for each column\n mode_values = {}\n for column in df.columns:\n try:\n mode_values[column] = df[column].mode().iloc[0]\n except IndexError:\n mode_values[column] = None # Handle cases with no mode\n\n # Calculate the percentage of most frequent values\n mode_rates = {}\n for column in df.columns:\n if column in mode_values and mode_values[column] is not None:\n mode_value = mode_values[column]\n mode_rate = (df[column] == mode_value).sum() / len(df)\n mode_rates[column] = mode_rate\n else:\n mode_rates[column] = None\n\n # Convert the most common value and its percentage into a data frame\n df_mode = pd.DataFrame(list(mode_values.items()), columns=['column_name', 'mode'])\n df_mode['rate mode'] = df_mode['column_name'].map(mode_rates)\n\n # Custom statistics calculation without using df.describe for basic stats\n def calculate_custom_stats(series):\n q1 = series.quantile(0.25)\n q3 = series.quantile(0.75)\n iqr = q3 - q1\n return pd.Series({\n 'count': series.count(),\n 'mean': series.mean(),\n 'std': series.std(),\n # 'q1-1.5*QTR': q1 - 1.5 * iqr,\n '0%': series.quantile(0),\n '25%': q1,\n '50%': series.median(),\n '75%': q3,\n '100%': series.quantile(1),\n 'lower_bound': series.mean() - 2 * series.std(),\n 'upper_bound': series.mean() + 2 * series.std()\n # 'q3+1.5*QTR': q3 + 1.5 * iqr\n })\n\n df_numeric = df.select_dtypes(include=np.number)\n custom_description = df_numeric.apply(calculate_custom_stats).T.reset_index(names='column_name')\n\n # process_three:Combining DataFrames and adding outlier rates, skewness, and kurtosis:\n data_inspection = pd.merge(data_inspection, custom_description, how='left', on='column_name')\n data_inspection = pd.merge(data_inspection, df_mode, how='left', on='column_name')\n\n # # The 'description' merge is no longer needed as 'custom_description' provides the stats\n # data_inspection = pd.merge(data_inspection, description, how='left', on='column_name')\n\n outliers_rate = check_if_outliers(df)\n outliers_rate = pd.DataFrame(list(outliers_rate.items()), columns=['column_name', 'outliers_rate']).reset_index(drop=True)\n data_inspection = pd.merge(data_inspection, outliers_rate, how='left', on='column_name')\n\n # skewness and kurtosis\n skew = df.select_dtypes(include=np.number).skew().reset_index().rename(columns={'index': 'column_name', 0: 'skewness'})\n kurt = df.select_dtypes(include=np.number).kurt().reset_index().rename(columns={'index': 'column_name', 0: 'kurtosis'})\n\n data_inspection = pd.merge(data_inspection, skew, how='left', on='column_name')\n data_inspection = pd.merge(data_inspection, kurt, how='left', on='column_name')\n\n # Calculating correlation coefficients (numeric columns only)\n df_numeric = df.select_dtypes(include=np.number)\n if not df_numeric.empty and df_numeric.shape[1] > 1:\n try:\n correlation_matrix = df_numeric.corr(numeric_only=True)\n\n # Assuming the last column is the target variable for correlation\n target_column = df_numeric.columns[-1]\n if target_column in correlation_matrix.index:\n target_corr = pd.DataFrame({'column_name': correlation_matrix.index, 'target correlation': correlation_matrix[target_column]})\n data_inspection = pd.merge(data_inspection, target_corr, how='left', on='column_name')\n else:\n data_inspection['target correlation'] = np.nan\n except Exception as e:\n print(f\"Error calculating correlation: {e}\")\n data_inspection['target correlation'] = np.nan\n else:\n data_inspection['target correlation'] = np.nan\n\n # process_four:Adding more column information and example data\n # --- MODIFIED PART START ---\n data_examples = {}\n for col in df.columns:\n non_null_values = df[col].dropna()\n if not non_null_values.empty:\n # Randomly select a non-null value\n example_value = non_null_values.sample(n=1).iloc[0]\n data_examples[col] = str(example_value).replace('\\n', '
')\n else:\n data_examples[col] = None # Or an empty string, depending on preference\n\n # process_four:Adding more column information and example data\n data_inspection_else = pd.DataFrame({\n 'column_name': df.columns,\n 'flag_or_not': df.apply(check_if_binary),\n 'columns_details': None,\n 'remarks': None,\n 'trigger': None,\n 'dataset_name': None,\n 'existence_of_table_definition': None,\n 'data_exmaple': pd.Series(data_examples) # Assign the generated examples\n })\n\n data_inspection = pd.merge(data_inspection, data_inspection_else, how='left', on='column_name')\n\n # visualization `data_inspection`\n # blue → green → yellow\n styled_columns = data_inspection.select_dtypes(include=np.number).columns\n if not styled_columns.empty:\n data_inspection_styled = data_inspection.style.background_gradient(cmap='viridis', subset=pd.IndexSlice[:, styled_columns])\n else:\n data_inspection_styled = data_inspection\n\n return data_inspection_styled\n","metadata":{"papermill":{"duration":0.024529,"end_time":"2025-06-11T13:37:19.610959","exception":false,"start_time":"2025-06-11T13:37:19.586430","status":"completed"},"tags":[],"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T13:45:58.266176Z","iopub.execute_input":"2025-07-09T13:45:58.266919Z","iopub.status.idle":"2025-07-09T13:45:58.281492Z","shell.execute_reply.started":"2025-07-09T13:45:58.266897Z","shell.execute_reply":"2025-07-09T13:45:58.280867Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"###
df_train","metadata":{"papermill":{"duration":0.007452,"end_time":"2025-06-11T13:37:19.626167","exception":false,"start_time":"2025-06-11T13:37:19.618715","status":"completed"},"tags":[]}},{"cell_type":"code","source":"data_inspection_styled = data_inspection(df_articles)\ndisplay(data_inspection_styled)\n# data_inspection.to_csv('data_inspection.csv', index = 'false') # Saving the dataframe if necessary","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T13:44:13.538907Z","iopub.execute_input":"2025-07-09T13:44:13.539478Z","iopub.status.idle":"2025-07-09T13:44:14.634234Z","shell.execute_reply.started":"2025-07-09T13:44:13.539458Z","shell.execute_reply":"2025-07-09T13:44:14.633502Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"data_inspection_styled = data_inspection(df_customers)\ndisplay(data_inspection_styled)\n# data_inspection.to_csv('data_inspection.csv', index = 'false') # Saving the dataframe if necessary","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T13:44:14.747126Z","iopub.execute_input":"2025-07-09T13:44:14.747389Z","iopub.status.idle":"2025-07-09T13:44:23.066445Z","shell.execute_reply.started":"2025-07-09T13:44:14.747371Z","shell.execute_reply":"2025-07-09T13:44:23.065795Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"data_inspection_styled = data_inspection(df_train)\ndisplay(data_inspection_styled)\n# data_inspection.to_csv('data_inspection.csv', index = 'false') # Saving the dataframe if necessary","metadata":{"papermill":{"duration":2.269356,"end_time":"2025-06-11T13:37:21.959848","exception":false,"start_time":"2025-06-11T13:37:19.690492","status":"completed"},"tags":[],"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T13:46:02.741119Z","iopub.execute_input":"2025-07-09T13:46:02.741384Z","iopub.status.idle":"2025-07-09T13:47:19.208493Z","shell.execute_reply.started":"2025-07-09T13:46:02.741364Z","shell.execute_reply":"2025-07-09T13:47:19.207771Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"display(len(df_articles) == len(df_articles.drop_duplicates()))\ndisplay(len(df_customers) == len(df_customers.drop_duplicates()))\ndisplay(len(df_train) == len(df_train.drop_duplicates()))","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T13:57:04.828734Z","iopub.execute_input":"2025-07-09T13:57:04.829504Z","iopub.status.idle":"2025-07-09T13:57:23.645059Z","shell.execute_reply.started":"2025-07-09T13:57:04.829474Z","shell.execute_reply":"2025-07-09T13:57:23.644423Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## df_articles","metadata":{}},{"cell_type":"code","source":"df_articles.head()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:04:41.087652Z","iopub.execute_input":"2025-07-09T14:04:41.087932Z","iopub.status.idle":"2025-07-09T14:04:41.100946Z","shell.execute_reply.started":"2025-07-09T14:04:41.087913Z","shell.execute_reply":"2025-07-09T14:04:41.100404Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# List of columns to visualize\ncolumns_to_visualize = [\n 'prod_name',\n 'product_type_name',\n 'product_group_name',\n 'department_name',\n 'index_name',\n 'index_group_name',\n 'section_name',\n 'garment_group_name',\n 'detail_desc'\n]\n\n# Loop through each column and create a bar plot\nfor col_name in columns_to_visualize:\n # Get the top 50 values for the current column\n # value_counts()の実行前にカテゴリカル型に変換することで、\n # 欠損値（NaN）がある場合にエラーが発生するのを防ぎ、\n # 欠損値をカウントから除外できます。\n top_50_values = df_articles[col_name].astype(str).value_counts().head(50)\n\n # Create a new figure for each plot\n plt.figure(figsize=(12, max(6, len(top_50_values) * 0.4))) # Adjust height based on number of bars\n\n # Create the bar plot\n sns.barplot(x=top_50_values.values, y=top_50_values.index, palette='viridis')\n\n # Set title and labels dynamically\n plt.title(f'Top 50 {col_name.replace(\"_\", \" \").title()} by Count', fontsize=16)\n plt.xlabel('Count', fontsize=12)\n plt.ylabel(col_name.replace(\"_\", \" \").title(), fontsize=12)\n\n # Adjust layout\n plt.tight_layout()\n\n # Display the plot\n plt.show()\n\nprint(\"All plots displayed successfully.\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:11:43.893817Z","iopub.execute_input":"2025-07-09T14:11:43.894499Z","iopub.status.idle":"2025-07-09T14:11:48.082232Z","shell.execute_reply.started":"2025-07-09T14:11:43.894476Z","shell.execute_reply":"2025-07-09T14:11:48.081530Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## df_customers","metadata":{}},{"cell_type":"code","source":"df_customers.head()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:02:27.629782Z","iopub.execute_input":"2025-07-09T14:02:27.630321Z","iopub.status.idle":"2025-07-09T14:02:27.640185Z","shell.execute_reply.started":"2025-07-09T14:02:27.630300Z","shell.execute_reply":"2025-07-09T14:02:27.639431Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# List of columns to visualize\ncolumns_to_visualize = [\n 'FN',\n 'Active',\n 'club_member_status',\n 'fashion_news_frequency',\n 'age'\n]\n\n# Loop through each column and create a bar plot\nfor col_name in columns_to_visualize:\n # Get the top 50 values for the current column\n # value_counts()の実行前にカテゴリカル型に変換することで、\n # 欠損値（NaN）がある場合にエラーが発生するのを防ぎ、\n # 欠損値をカウントから除外できます。\n top_50_values = df_customers[col_name].astype(str).value_counts().head(50)\n\n # Create a new figure for each plot\n plt.figure(figsize=(12, max(6, len(top_50_values) * 0.4))) # Adjust height based on number of bars\n\n # Create the bar plot\n sns.barplot(x=top_50_values.values, y=top_50_values.index, palette='viridis')\n\n # Set title and labels dynamically\n plt.title(f'Top 50 {col_name.replace(\"_\", \" \").title()} by Count', fontsize=16)\n plt.xlabel('Count', fontsize=12)\n plt.ylabel(col_name.replace(\"_\", \" \").title(), fontsize=12)\n\n # Adjust layout\n plt.tight_layout()\n\n # Display the plot\n plt.show()\n\nprint(\"All plots displayed successfully.\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:14:14.247544Z","iopub.execute_input":"2025-07-09T14:14:14.248291Z","iopub.status.idle":"2025-07-09T14:14:17.851073Z","shell.execute_reply.started":"2025-07-09T14:14:14.248270Z","shell.execute_reply":"2025-07-09T14:14:17.850395Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## df_train","metadata":{}},{"cell_type":"code","source":"display(len(df_train))\ndf_train['t_dat'] = pd.to_datetime(df_train['t_dat'], errors='coerce')\ndf_train.head()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:21:38.684049Z","iopub.execute_input":"2025-07-09T14:21:38.684744Z","iopub.status.idle":"2025-07-09T14:21:41.774073Z","shell.execute_reply.started":"2025-07-09T14:21:38.684719Z","shell.execute_reply":"2025-07-09T14:21:41.773457Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"top_50_detail_desc = df_train['t_dat'].value_counts().head(50)\n\nplt.figure(figsize=(12, 10))\nsns.barplot(x=top_50_detail_desc.values, y=top_50_detail_desc.index, palette='viridis')\n\nplt.title('Top 50 t_dat by Count', fontsize=16)\nplt.xlabel('Count', fontsize=12)\nplt.ylabel('t_dat', fontsize=12)\nplt.tight_layout()\nplt.show()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:23:25.566639Z","iopub.execute_input":"2025-07-09T14:23:25.567100Z","iopub.status.idle":"2025-07-09T14:23:26.274102Z","shell.execute_reply.started":"2025-07-09T14:23:25.567077Z","shell.execute_reply":"2025-07-09T14:23:26.273400Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"df_train['t_dat'].max()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:24:26.283061Z","iopub.execute_input":"2025-07-09T14:24:26.283587Z","iopub.status.idle":"2025-07-09T14:24:26.364491Z","shell.execute_reply.started":"2025-07-09T14:24:26.283562Z","shell.execute_reply":"2025-07-09T14:24:26.363889Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# 最新の日付を取得\nmax_date = df_train['t_dat'].max()\nthree_months_ago = max_date - pd.DateOffset(months=3)\ndf_last_3_months = df_train[df_train['t_dat'] >= three_months_ago]\n\ndisplay(three_months_ago), display(max_date)\ndf_last_3_months","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:27:37.071178Z","iopub.execute_input":"2025-07-09T14:27:37.071974Z","iopub.status.idle":"2025-07-09T14:27:37.456838Z","shell.execute_reply.started":"2025-07-09T14:27:37.071941Z","shell.execute_reply":"2025-07-09T14:27:37.456195Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"top_50_detail_desc = df_last_3_months['t_dat'].value_counts().head(50)\n\nplt.figure(figsize=(12, 10))\nsns.barplot(x=top_50_detail_desc.values, y=top_50_detail_desc.index, palette='viridis')\n\nplt.title('Top 50 t_dat by Count', fontsize=16)\nplt.xlabel('Count', fontsize=12)\nplt.ylabel('t_dat', fontsize=12)\nplt.tight_layout()\nplt.show()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:26:56.516015Z","iopub.execute_input":"2025-07-09T14:26:56.516307Z","iopub.status.idle":"2025-07-09T14:26:57.211550Z","shell.execute_reply.started":"2025-07-09T14:26:56.516286Z","shell.execute_reply":"2025-07-09T14:26:57.210767Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"df_merge_trainsaction_customer = pd.merge(df_last_3_months, df_customers, how = 'inner', on = 'customer_id')\ndf_merge_trainsaction_customer","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:29:07.381119Z","iopub.execute_input":"2025-07-09T14:29:07.381606Z","iopub.status.idle":"2025-07-09T14:29:09.343901Z","shell.execute_reply.started":"2025-07-09T14:29:07.381583Z","shell.execute_reply":"2025-07-09T14:29:09.343155Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# List of columns to visualize\ncolumns_to_visualize = [\n 'FN',\n 'Active',\n 'club_member_status',\n 'fashion_news_frequency',\n 'age'\n]\n\n# Loop through each column and create a bar plot\nfor col_name in columns_to_visualize:\n # Get the top 50 values for the current column\n # value_counts()の実行前にカテゴリカル型に変換することで、\n # 欠損値（NaN）がある場合にエラーが発生するのを防ぎ、\n # 欠損値をカウントから除外できます。\n top_50_values = df_merge_trainsaction_customer[col_name].astype(str).value_counts().head(50)\n\n # Create a new figure for each plot\n plt.figure(figsize=(12, max(6, len(top_50_values) * 0.4))) # Adjust height based on number of bars\n\n # Create the bar plot\n sns.barplot(x=top_50_values.values, y=top_50_values.index, palette='viridis')\n\n # Set title and labels dynamically\n plt.title(f'Top 50 {col_name.replace(\"_\", \" \").title()} by Count', fontsize=16)\n plt.xlabel('Count', fontsize=12)\n plt.ylabel(col_name.replace(\"_\", \" \").title(), fontsize=12)\n\n # Adjust layout\n plt.tight_layout()\n\n # Display the plot\n plt.show()\n\nprint(\"All plots displayed successfully.\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:29:31.933690Z","iopub.execute_input":"2025-07-09T14:29:31.934398Z","iopub.status.idle":"2025-07-09T14:29:40.047970Z","shell.execute_reply.started":"2025-07-09T14:29:31.934373Z","shell.execute_reply":"2025-07-09T14:29:40.047225Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"df_merge_trainsaction_article = pd.merge(df_last_3_months, df_articles, how = 'inner', on = 'article_id')\ndf_merge_trainsaction_article","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:30:47.321975Z","iopub.execute_input":"2025-07-09T14:30:47.322222Z","iopub.status.idle":"2025-07-09T14:30:49.055858Z","shell.execute_reply.started":"2025-07-09T14:30:47.322205Z","shell.execute_reply":"2025-07-09T14:30:49.055109Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# List of columns to visualize\ncolumns_to_visualize = [\n 'prod_name',\n 'product_type_name',\n 'product_group_name',\n 'department_name',\n 'index_name',\n 'index_group_name',\n 'section_name',\n 'garment_group_name',\n 'detail_desc'\n]\n\n# Loop through each column and create a bar plot\nfor col_name in columns_to_visualize:\n # Get the top 50 values for the current column\n # value_counts()の実行前にカテゴリカル型に変換することで、\n # 欠損値（NaN）がある場合にエラーが発生するのを防ぎ、\n # 欠損値をカウントから除外できます。\n top_50_values = df_merge_trainsaction_article[col_name].astype(str).value_counts().head(50)\n\n # Create a new figure for each plot\n plt.figure(figsize=(12, max(6, len(top_50_values) * 0.4))) # Adjust height based on number of bars\n\n # Create the bar plot\n sns.barplot(x=top_50_values.values, y=top_50_values.index, palette='viridis')\n\n # Set title and labels dynamically\n plt.title(f'Top 50 {col_name.replace(\"_\", \" \").title()} by Count', fontsize=16)\n plt.xlabel('Count', fontsize=12)\n plt.ylabel(col_name.replace(\"_\", \" \").title(), fontsize=12)\n\n # Adjust layout\n plt.tight_layout()\n\n # Display the plot\n plt.show()\n\nprint(\"All plots displayed successfully.\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-07-09T14:31:35.771218Z","iopub.execute_input":"2025-07-09T14:31:35.771865Z","iopub.status.idle":"2025-07-09T14:31:42.534658Z","shell.execute_reply.started":"2025-07-09T14:31:35.771840Z","shell.execute_reply":"2025-07-09T14:31:42.533923Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"","metadata":{"trusted":true},"outputs":[],"execution_count":null}]}