data_storytelling.py

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from wordcloud import WordCloud

class DataStoryteller:
    def __init__(self):
        pass

    def generate_story(self, data):
        story = "Data Story:\n\n"
        
        # Basic statistics
        story += self._generate_basic_stats(data)
        
        # Correlation analysis
        story += self._generate_correlation_analysis(data)
        
        # Trend analysis
        story += self._generate_trend_analysis(data)
        
        # Distribution analysis
        story += self._generate_distribution_analysis(data)
        
        return story

    def _generate_basic_stats(self, data):
        stats = data.describe()
        text = "Basic Statistics:\n"
        for column in stats.columns:
            text += f"\n{column}:\n"
            text += f"  Mean: {stats[column]['mean']:.2f}\n"
            text += f"  Median: {data[column].median():.2f}\n"
            text += f"  Min: {stats[column]['min']:.2f}\n"
            text += f"  Max: {stats[column]['max']:.2f}\n"
        return text

    def _generate_correlation_analysis(self, data):
        numeric_data = data.select_dtypes(include=[np.number])
        corr_matrix = numeric_data.corr()
        
        text = "\nCorrelation Analysis:\n"
        for i in range(len(corr_matrix.columns)):
            for j in range(i+1, len(corr_matrix.columns)):
                col1, col2 = corr_matrix.columns[i], corr_matrix.columns[j]
                corr = corr_matrix.loc[col1, col2]
                if abs(corr) > 0.5:
                    text += f"  Strong correlation between {col1} and {col2}: {corr:.2f}\n"
        return text

    def _generate_trend_analysis(self, data):
        text = "\nTrend Analysis:\n"
        for column in data.select_dtypes(include=[np.number]).columns:
            trend = np.polyfit(range(len(data)), data[column], 1)[0]
            if trend > 0:
                text += f"  {column} shows an increasing trend.\n"
            elif trend < 0:
                text += f"  {column} shows a decreasing trend.\n"
            else:
                text += f"  {column} shows no significant trend.\n"
        return text

    def _generate_distribution_analysis(self, data):
        text = "\nDistribution Analysis:\n"
        for column in data.select_dtypes(include=[np.number]).columns:
            skewness = data[column].skew()
            if abs(skewness) < 0.5:
                text += f"  {column} is approximately symmetrically distributed.\n"
            elif skewness > 0:
                text += f"  {column} is right-skewed.\n"
            else:
                text += f"  {column} is left-skewed.\n"
        return text

    def generate_word_cloud(self, data, text_column):
        text = " ".join(data[text_column].astype(str))
        wordcloud = WordCloud(width=800, height=400, background_color='white').generate(text)
        
        plt.figure(figsize=(10, 5))
        plt.imshow(wordcloud, interpolation='bilinear')
        plt.axis('off')
        plt.title('Word Cloud')
        
        return plt

    def generate_summary_dashboard(self, data):
        fig, axs = plt.subplots(2, 2, figsize=(20, 15))
        
        # Histogram
        sns.histplot(data=data, x=data.select_dtypes(include=[np.number]).columns[0], ax=axs[0, 0])
        axs[0, 0].set_title('Distribution of ' + data.select_dtypes(include=[np.number]).columns[0])
        
        # Scatter plot
        sns.scatterplot(data=data, x=data.select_dtypes(include=[np.number]).columns[0], 
                        y=data.select_dtypes(include=[np.number]).columns[1], ax=axs[0, 1])
        axs[0, 1].set_title('Scatter Plot')
        
        # Box plot
        sns.boxplot(data=data, y=data.select_dtypes(include=[np.number]).columns[0], ax=axs[1, 0])
        axs[1, 0].set_title('Box Plot')
        
        # Correlation heatmap
        sns.heatmap(data.select_dtypes(include=[np.number]).corr(), annot=True, cmap='coolwarm', ax=axs[1, 1])
        axs[1, 1].set_title('Correlation Heatmap')
        
        plt.tight_layout()
        return fig