old/deprecated.py

with gr.Tab("3️⃣ Correlation Explorer"):
        col_select = gr.Dropdown(choices=categorical_cols.tolist(), label="Select Column")
        col1 = gr.Dropdown(choices=df.columns.tolist(), label="Column 1")
        col2 = gr.Dropdown(choices=df.columns.tolist(), label="Column 2")
        corr_btn = gr.Button("Compute Correlation")
        corr_text = gr.Textbox(label="Correlation Result")
        confusion_out = gr.Dataframe(label="Categorical Crosstab (if applicable)")
        corr_btn.click(compute_correlation, inputs=[col1, col2], outputs=[corr_text, confusion_out])


def compute_correlation(col1, col2):
    c1 = df[col1]
    c2 = df[col2]

    # Case 1: numeric vs numeric
    if col1 in numeric_cols and col2 in numeric_cols:
        corr = c1.corr(c2)
        return f"Pearson Correlation = {corr:.4f}", None

    # Case 2: categorical vs categorical → Cramér’s V
    if col1 in categorical_cols and col2 in categorical_cols:
        confusion = pd.crosstab(c1, c2)
        v = cramers_v(confusion)
        return f"Cramér’s V = {v:.4f}", confusion

    # Case 3: keyword frequency vs numeric/categorical
    # Convert col1 or col2 (if comma-separated) into frequency counts
    def keyword_expand(col):
        return df[col].dropna().astype(str).str.split(',').explode().str.strip()

    if col1 in categorical_cols:
        k = keyword_expand(col1)
        k_counts = k.value_counts()
        df_k = df.assign(**{f"{col1}_KEYWORD_COUNTS": df[col1].fillna("").apply(
            lambda x: sum([k_counts.get(i.strip(), 0) for i in x.split(',') if i.strip()])
        )})
        c1 = df_k[f"{col1}_KEYWORD_COUNTS"]

    if col2 in categorical_cols:
        k = keyword_expand(col2)
        k_counts = k.value_counts()
        df_k = df.assign(**{f"{col2}_KEYWORD_COUNTS": df[col2].fillna("").apply(
            lambda x: sum([k_counts.get(i.strip(), 0) for i in x.split(',') if i.strip()])
        )})
        c2 = df_k[f"{col2}_KEYWORD_COUNTS"]

    corr = c1.corr(c2)
    return f"Keyword-Frequency Based Correlation = {corr:.4f}", None