insightflow-ai 1.0.0__tar.gz

insightflow_ai-1.0.0/InsightFlow/__init__.py

@@ -0,0 +1,51 @@
+import pandas as pd
+from .autofix import run_autofix, run_autofix_file, set_hf_token as autofix_set_token
+from .autoeda import run_autoeda, run_autoeda_file, set_hf_token as autoeda_set_token
+from .automodel import run_automodel, run_automodel_file, set_hf_token as automodel_set_token
+from .autoreport import run_autoreport, set_hf_token as autoreport_set_token
+
+def _sync_tokens(hf_token: str):
+    """Sets the token across all modules"""
+    autofix_set_token(hf_token)
+    autoeda_set_token(hf_token)
+    automodel_set_token(hf_token)
+    autoreport_set_token(hf_token)
+
+def autofix(data, hf_token: str, save_path="cleaned_data.csv"):
+    """
+    Cleans messy data using AI.
+    `data` can be a string (file path) or a pandas DataFrame.
+    """
+    _sync_tokens(hf_token)
+    if isinstance(data, str):
+        return run_autofix_file(data, save_path=save_path)
+    return run_autofix(data, save_path=save_path)
+
+def autoeda(data, hf_token: str, save_path="eda_report.html"):
+    """
+    Generates an Exploratory Data Analysis report.
+    `data` can be a string (file path) or a pandas DataFrame.
+    """
+    _sync_tokens(hf_token)
+    if isinstance(data, str):
+        return run_autoeda_file(data, save_path=save_path)
+    return run_autoeda(data, save_path=save_path)
+
+def automodel(data, hf_token: str, target_col=None, task_type="auto", save_path="model_report.html"):
+    """
+    Trains multiple ML models and selects the best one.
+    `data` can be a string (file path) or a pandas DataFrame.
+    """
+    _sync_tokens(hf_token)
+    if isinstance(data, str):
+        return run_automodel_file(data, target_col=target_col, task_type=task_type, save_path=save_path)
+    return run_automodel(data, target_col=target_col, task_type=task_type, save_path=save_path)
+
+def autoreport(eda_results: dict, eda_insights: dict, model_results: dict, model_insights: dict, target_col: str, hf_token: str, save_path="business_report.html"):
+    """
+    Compiles an advanced C-Level business report using all previous AI insights.
+    """
+    _sync_tokens(hf_token)
+    return run_autoreport(eda_results, eda_insights, model_results, model_insights, target_col, save_path=save_path)
+
+__all__ = ["autofix", "autoeda", "automodel", "autoreport"]

insightflow_ai-1.0.0/InsightFlow/autoeda.py

@@ -0,0 +1,75 @@
+import os
+import pandas as pd
+from .autoeda_agent import AutoEDAAgent
+
+
+_hf_token = None
+
+
+def set_hf_token(token: str):
+    global _hf_token
+    _hf_token = token
+    os.environ["HF_TOKEN"] = token
+    print("HF token set successfully.")
+
+
+def get_hf_token():
+    global _hf_token
+    if _hf_token:
+        return _hf_token
+    return os.environ.get("HF_TOKEN", None)
+
+
+def run_autoeda(df: pd.DataFrame, save_path: str = "eda_report.html"):
+    """
+    Runs AI-powered EDA on a DataFrame and generates an HTML report.
+
+    Args:
+        df: pandas DataFrame to analyze
+        save_path: path to save the HTML report
+
+    Returns:
+        Tuple of (eda_results dict, ai_insights dict)
+    """
+    token = get_hf_token()
+    if not token:
+        raise RuntimeError("HF token not set. Call set_hf_token('your_token') first.")
+    agent = AutoEDAAgent(token)
+    return agent.run(df, save_path=save_path)
+
+
+def run_autoeda_file(file_path: str, save_path: str = None):
+    """
+    Loads a CSV/Excel file and runs AI-powered EDA.
+
+    Args:
+        file_path: path to CSV or Excel file
+        save_path: path to save the HTML report
+
+    Returns:
+        Tuple of (eda_results dict, ai_insights dict)
+    """
+    token = get_hf_token()
+    if not token:
+        raise RuntimeError("HF token not set. Call set_hf_token('your_token') first.")
+    agent = AutoEDAAgent(token)
+    return agent.run_file(file_path, save_path=save_path)
+
+
+if __name__ == "__main__":
+    import sys
+
+    print("=" * 50)
+    print("  InsightFlow Auto EDA")
+    print("=" * 50)
+
+    token = input("\nEnter your HuggingFace API token: ").strip()
+    set_hf_token(token)
+
+    file_path = input("Enter the path to your data file: ").strip()
+    if not os.path.exists(file_path):
+        print(f"Error: File '{file_path}' not found.")
+        sys.exit(1)
+
+    run_autoeda_file(file_path)
+    print("\nDone!")

insightflow_ai-1.0.0/InsightFlow/autoeda_agent.py

@@ -0,0 +1,410 @@
+import os
+import json
+import pandas as pd
+import numpy as np
+from openai import OpenAI
+from .eda import run_full_eda, dataset_overview, column_stats
+from .report_generator import build_report
+
+
+class AutoEDAAgent:
+    def __init__(self, hf_token: str, model: str = "zai-org/GLM-5.2:novita"):
+        self.client = OpenAI(base_url="https://router.huggingface.co/v1", api_key=hf_token)
+        self.model = model
+
+    def _ask_llm(self, eda_results):
+        overview = eda_results["overview"]
+        col_info = {}
+        for k, v in eda_results.get("column_stats", {}).items():
+            col_info[k] = {kk: vv for kk, vv in v.items() if kk != "top5"}
+
+        dist_info = {}
+        for k, v in eda_results.get("distributions", {}).items():
+            dist_info[k] = {kk: vv for kk, vv in v.items() if kk != "histogram"}
+
+        corr_matrix = eda_results.get("correlation", {}).get("matrix")
+        if hasattr(corr_matrix, "to_dict"):
+            eda_results["correlation"]["matrix"] = corr_matrix.to_dict()
+
+        prompt = f"""You are an EDA expert. Analyze these dataset results and provide insights.
+
+OVERVIEW: {json.dumps(overview)}
+COLUMN STATS: {json.dumps(col_info, default=str)}
+MISSING: {json.dumps(eda_results.get("missing", {}))}
+OUTLIERS: {json.dumps(eda_results.get("outliers", {}))}
+STRONG CORRELATIONS: {json.dumps(eda_results.get("correlation", {}).get("strong_pairs", []))}
+CATEGORICAL: {json.dumps(eda_results.get("categorical", {}), default=str)}
+DISTRIBUTIONS: {json.dumps(dist_info)}
+
+Return a JSON object with these keys:
+- "summary": 2-3 sentence overall summary
+- "key_findings": list of 5-8 important findings (strings)
+- "warnings": list of data quality warnings (strings)
+- "recommendations": list of 3-5 actionable recommendations (strings)
+- "target_suggestion": which column could be a good prediction target and why (string)
+
+Respond ONLY with valid JSON. No markdown, no explanation."""
+
+        response = self.client.chat.completions.create(
+            model=self.model,
+            messages=[{"role": "user", "content": prompt}],
+            temperature=0.1,
+        )
+        raw = response.choices[0].message.content.strip()
+        if raw.startswith("```"):
+            lines = [l for l in raw.split("\n") if not l.startswith("```")]
+            raw = "\n".join(lines)
+        return json.loads(raw)
+
+    def _build_html(self, df, eda_results, ai_insights, save_path):
+        import json
+        import datetime
+        now = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+
+        # Safely dump data to inject into Javascript
+        def safe_json(obj):
+            return json.dumps(obj, default=str).replace("</", "<\\/")
+
+        eda_json = safe_json(eda_results)
+        ai_json = safe_json(ai_insights)
+
+        html = f"""<!DOCTYPE html>
+<html lang="en">
+<head>
+<meta charset="UTF-8"><meta name="viewport" content="width=device-width,initial-scale=1.0">
+<title>InsightFlow AutoEDA Pro</title>
+<link href="https://fonts.googleapis.com/css2?family=Outfit:wght@300;400;500;600;700;800&family=JetBrains+Mono:wght@400;700&display=swap" rel="stylesheet">
+<script src="https://cdn.jsdelivr.net/npm/chart.js@4.4.0/dist/chart.umd.min.js"></script>
+<style>
+:root {{
+    --bg: #09090b; --surface: #18181b; --surface-hover: #27272a;
+    --border: #3f3f46; --border-light: #27272a;
+    --text: #f4f4f5; --text-muted: #a1a1aa;
+    --primary: #3b82f6; --primary-glow: rgba(59,130,246,0.3);
+    --success: #10b981; --warning: #f59e0b; --danger: #ef4444; --accent: #8b5cf6;
+}}
+* {{ margin:0; padding:0; box-sizing:border-box; }}
+body {{ font-family:'Outfit',sans-serif; background:var(--bg); color:var(--text); line-height:1.6; padding-bottom:4rem; }}
+.bg-grid {{ position:fixed; inset:0; background-image:linear-gradient(var(--border-light) 1px,transparent 1px),linear-gradient(90deg,var(--border-light) 1px,transparent 1px); background-size:40px 40px; pointer-events:none; opacity:0.3; z-index:-1; }}
+.bg-glow {{ position:fixed; top: -20%; left: -10%; width: 50%; height: 50%; background: radial-gradient(circle, var(--primary-glow) 0%, transparent 60%); filter: blur(100px); z-index:-1; pointer-events:none; }}
+.bg-glow-2 {{ position:fixed; bottom: -20%; right: -10%; width: 50%; height: 50%; background: radial-gradient(circle, rgba(139,92,246,0.2) 0%, transparent 60%); filter: blur(100px); z-index:-1; pointer-events:none; }}
+.container {{ max-width:1400px; margin:0 auto; padding:2rem; }}
+
+header {{ text-align:center; padding:4rem 2rem; background:var(--surface); border-radius:24px; border:1px solid var(--border); position:relative; overflow:hidden; margin-bottom:2rem; box-shadow: 0 25px 50px -12px rgba(0,0,0,0.5); }}
+header::after {{ content:''; position:absolute; inset:0; background: linear-gradient(180deg, transparent, rgba(0,0,0,0.4)); pointer-events:none; }}
+h1 {{ font-size:3.5rem; font-weight:800; letter-spacing:-0.05em; margin-bottom:0.5rem; background:linear-gradient(to right, #60a5fa, #c084fc); -webkit-background-clip:text; -webkit-text-fill-color:transparent; }}
+.subtitle {{ font-size:1.1rem; color:var(--text-muted); font-weight:400; }}
+.timestamp {{ font-size:0.85rem; color:#52525b; margin-top:1rem; font-family:'JetBrains Mono',monospace; }}
+
+.grid-metrics {{ display:grid; grid-template-columns:repeat(auto-fit,minmax(200px,1fr)); gap:1rem; margin-bottom:2rem; }}
+.card {{ background:rgba(24,24,27,0.7); backdrop-filter:blur(16px); border:1px solid var(--border); border-radius:20px; padding:1.5rem; transition:transform 0.3s, border-color 0.3s; }}
+.card:hover {{ transform:translateY(-5px); border-color:var(--primary); }}
+.metric-val {{ font-size:2.5rem; font-weight:800; color:var(--text); line-height:1; margin-bottom:0.25rem; }}
+.metric-lbl {{ font-size:0.8rem; color:var(--text-muted); text-transform:uppercase; letter-spacing:0.1em; font-weight:600; }}
+
+.section-title {{ font-size:1.75rem; font-weight:700; margin-bottom:1.5rem; display:flex; align-items:center; gap:0.5rem; margin-top:3rem; }}
+.section-title span {{ color:var(--primary); }}
+
+.ai-box {{ background:linear-gradient(145deg, rgba(24,24,27,0.9), rgba(9,9,11,0.9)); border:1px solid rgba(139,92,246,0.3); position:relative; overflow:hidden; }}
+.ai-box::before {{ content:''; position:absolute; top:0; left:0; right:0; height:2px; background:linear-gradient(90deg, #3b82f6, #8b5cf6, #ec4899); }}
+.ai-summary {{ font-size:1.1rem; color:#d4d4d8; margin-bottom:1.5rem; line-height:1.7; }}
+.insight-list {{ display:flex; flex-direction:column; gap:0.75rem; }}
+.insight-item {{ padding:1rem; border-radius:12px; background:rgba(255,255,255,0.03); border-left:4px solid var(--primary); display:flex; gap:1rem; align-items:flex-start; font-size:0.95rem; }}
+.insight-item.warn {{ border-color:var(--warning); background:rgba(245,158,11,0.05); }}
+.insight-item.rec {{ border-color:var(--success); background:rgba(16,185,129,0.05); }}
+.insight-item.target {{ border-color:var(--accent); background:rgba(139,92,246,0.1); padding:1.5rem; font-size:1.05rem; margin-top:1rem; }}
+.icon {{ font-size:1.25rem; line-height:1; }}
+
+.grid-charts {{ display:grid; grid-template-columns:repeat(auto-fit,minmax(400px,1fr)); gap:1.5rem; }}
+.chart-wrapper {{ height:300px; width:100%; position:relative; }}
+
+.table-wrap {{ overflow-x:auto; border-radius:16px; border:1px solid var(--border); background:rgba(24,24,27,0.5); backdrop-filter:blur(10px); }}
+table {{ width:100%; border-collapse:collapse; text-align:left; font-size:0.9rem; }}
+th, td {{ padding:1rem; border-bottom:1px solid var(--border-light); }}
+th {{ background:rgba(255,255,255,0.02); font-weight:600; color:var(--text-muted); text-transform:uppercase; letter-spacing:0.05em; font-size:0.75rem; }}
+tr:last-child td {{ border-bottom:none; }}
+tr:hover td {{ background:rgba(255,255,255,0.02); }}
+.mono {{ font-family:'JetBrains Mono',monospace; font-size:0.85rem; color:#94a3b8; }}
+
+.badge {{ padding:0.25rem 0.6rem; border-radius:99px; font-size:0.7rem; font-weight:700; text-transform:uppercase; letter-spacing:0.05em; }}
+.b-num {{ background:rgba(59,130,246,0.15); color:#60a5fa; }}
+.b-cat {{ background:rgba(168,85,247,0.15); color:#c084fc; }}
+.b-bool {{ background:rgba(245,158,11,0.15); color:#fbbf24; }}
+
+.corr-grid {{ display:flex; flex-direction:column; gap:0.25rem; font-family:'JetBrains Mono',monospace; font-size:0.8rem; overflow-x:auto; padding:1rem; }}
+.corr-row {{ display:flex; gap:0.25rem; }}
+.corr-cell {{ width:60px; height:40px; display:flex; align-items:center; justify-content:center; border-radius:6px; color:#fff; font-weight:600; flex-shrink:0; }}
+.corr-label {{ width:120px; justify-content:flex-end; padding-right:1rem; color:var(--text-muted); font-family:'Outfit',sans-serif; }}
+
+@media (max-width:768px) {{ .grid-charts {{ grid-template-columns:1fr; }} }}
+</style>
+</head>
+<body>
+<div class="bg-grid"></div>
+<div class="bg-glow"></div>
+<div class="bg-glow-2"></div>
+<div class="container">
+
+<header>
+    <h1>InsightFlow AutoEDA</h1>
+    <div class="subtitle" id="h-subtitle"></div>
+    <div class="timestamp">Generated: {now}</div>
+</header>
+
+<div class="grid-metrics" id="metrics-container"></div>
+
+<div class="card ai-box">
+    <div class="section-title" style="margin-top:0;"><span>✨</span> AI Insights</div>
+    <div class="ai-summary" id="ai-summary"></div>
+    <div class="insight-list" id="ai-insights"></div>
+</div>
+
+<h2 class="section-title"><span>📈</span> Auto Plots (Distributions & Frequencies)</h2>
+<div class="grid-charts" id="auto-plots"></div>
+
+<h2 class="section-title"><span>🔍</span> Correlation Matrix</h2>
+<div class="card" style="padding:0;">
+    <div class="corr-grid" id="corr-matrix"></div>
+</div>
+
+<h2 class="section-title"><span>📊</span> Dataset Statistics</h2>
+<div class="table-wrap">
+    <table id="stats-table">
+        <thead>
+            <tr><th>Feature</th><th>Type</th><th>Valid</th><th>Missing</th><th>Unique</th><th>Key Stats / Top Values</th></tr>
+        </thead>
+        <tbody></tbody>
+    </table>
+</div>
+
+</div>
+
+<script>
+const eda = {eda_json};
+const ai = {ai_json};
+
+// Format large numbers
+const fmt = (n) => new Intl.NumberFormat().format(n);
+
+// 1. Header & Metrics
+document.getElementById('h-subtitle').innerText = `${{fmt(eda.overview.rows)}} rows × ${{fmt(eda.overview.columns)}} columns | ${{eda.overview.memory_mb}} MB`;
+
+const metricsHtml = `
+    <div class="card"><div class="metric-val">${{fmt(eda.overview.rows)}}</div><div class="metric-lbl">Total Rows</div></div>
+    <div class="card"><div class="metric-val">${{fmt(eda.overview.columns)}}</div><div class="metric-lbl">Features</div></div>
+    <div class="card"><div class="metric-val">${{eda.overview.missing_pct}}%</div><div class="metric-lbl">Missing Data</div></div>
+    <div class="card"><div class="metric-val">${{fmt(eda.overview.duplicate_rows)}}</div><div class="metric-lbl">Duplicates</div></div>
+`;
+document.getElementById('metrics-container').innerHTML = metricsHtml;
+
+// 2. AI Insights
+document.getElementById('ai-summary').innerText = ai.summary || 'AI Analysis unavailable.';
+let insightsHtml = '';
+if(ai.key_findings) ai.key_findings.forEach(f => insightsHtml += `<div class="insight-item"><span class="icon">💡</span><div>${{f}}</div></div>`);
+if(ai.warnings) ai.warnings.forEach(w => insightsHtml += `<div class="insight-item warn"><span class="icon">⚠️</span><div>${{w}}</div></div>`);
+if(ai.recommendations) ai.recommendations.forEach(r => insightsHtml += `<div class="insight-item rec"><span class="icon">✅</span><div>${{r}}</div></div>`);
+if(ai.target_suggestion) insightsHtml += `<div class="insight-item target"><span class="icon">🎯</span><div><strong>Target Column Suggestion:</strong><br>${{ai.target_suggestion}}</div></div>`;
+document.getElementById('ai-insights').innerHTML = insightsHtml;
+
+// 3. Stats Table
+const tbody = document.querySelector('#stats-table tbody');
+Object.keys(eda.column_stats).forEach(col => {{
+    const stat = eda.column_stats[col];
+    const isNum = stat.dtype.includes('int') || stat.dtype.includes('float');
+    const isBool = stat.dtype.includes('bool');
+    const badgeClass = isNum ? 'b-num' : (isBool ? 'b-bool' : 'b-cat');
+    const badgeText = isNum ? 'NUM' : (isBool ? 'BOOL' : 'CAT');
+    
+    let extra = '';
+    if(isNum) extra = `μ = ${{stat.mean}} | σ = ${{stat.std}} | Range: [${{stat.min}} .. ${{stat.max}}]`;
+    else if(stat.top_value) extra = `Top: ${{stat.top_value}} (${{stat.top_freq}}x)`;
+
+    let missText = stat.missing > 0 ? `<span style="color:var(--danger)">${{stat.missing}} (${{stat.missing_pct}}%)</span>` : '0';
+
+    tbody.innerHTML += `
+        <tr>
+            <td style="font-weight:600; color:var(--text);">${{col}}</td>
+            <td><span class="badge ${{badgeClass}}">${{badgeText}}</span></td>
+            <td>${{stat.count}}</td>
+            <td>${{missText}}</td>
+            <td>${{stat.unique}}</td>
+            <td class="mono">${{extra}}</td>
+        </tr>
+    `;
+}});
+
+// 4. Auto Plots
+const plotsContainer = document.getElementById('auto-plots');
+let chartIndex = 0;
+
+// Colors for charts
+const colors = [
+    'rgba(59, 130, 246, 0.8)', 'rgba(139, 92, 246, 0.8)', 
+    'rgba(16, 185, 129, 0.8)', 'rgba(245, 158, 11, 0.8)',
+    'rgba(236, 72, 153, 0.8)'
+];
+
+// Numeric Distributions (Histograms)
+if(eda.distributions) {{
+    Object.keys(eda.distributions).forEach((col, i) => {{
+        const dist = eda.distributions[col];
+        if(!dist.histogram) return;
+        
+        const cid = `chart-${{chartIndex++}}`;
+        plotsContainer.innerHTML += `
+            <div class="card">
+                <h3 style="font-size:1rem; margin-bottom:1rem;">${{col}} <span style="font-size:0.75rem; color:var(--text-muted); font-weight:normal;">(${{dist.skew_label}})</span></h3>
+                <div class="chart-wrapper"><canvas id="${{cid}}"></canvas></div>
+            </div>
+        `;
+        
+        // Use timeout to allow DOM to render canvas elements first
+        setTimeout(() => {{
+            const ctx = document.getElementById(cid);
+            // Build labels from edges
+            const labels = [];
+            for(let j=0; j<dist.histogram.counts.length; j++){{
+                labels.push(`${{dist.histogram.edges[j].toFixed(1)}} - ${{dist.histogram.edges[j+1].toFixed(1)}}`);
+            }}
+            
+            new Chart(ctx, {{
+                type: 'bar',
+                data: {{
+                    labels: labels,
+                    datasets: [{{
+                        label: 'Count',
+                        data: dist.histogram.counts,
+                        backgroundColor: colors[i % colors.length],
+                        borderRadius: 4
+                    }}]
+                }},
+                options: {{
+                    responsive: true, maintainAspectRatio: false,
+                    plugins: {{ legend: {{ display: false }} }},
+                    scales: {{ 
+                        x: {{ ticks: {{ color: '#a1a1aa', maxTicksLimit: 6 }}, grid: {{ display: false }} }},
+                        y: {{ ticks: {{ color: '#a1a1aa' }}, grid: {{ color: 'rgba(255,255,255,0.05)' }} }}
+                    }}
+                }}
+            }});
+        }}, 100);
+    }});
+}}
+
+// Categorical Distributions (Doughnuts)
+if(eda.categorical) {{
+    Object.keys(eda.categorical).forEach((col, i) => {{
+        const cat = eda.categorical[col];
+        if(!cat.top_values) return;
+        
+        const cid = `chart-${{chartIndex++}}`;
+        plotsContainer.innerHTML += `
+            <div class="card">
+                <h3 style="font-size:1rem; margin-bottom:1rem;">${{col}} <span style="font-size:0.75rem; color:var(--text-muted); font-weight:normal;">(Top Values)</span></h3>
+                <div class="chart-wrapper"><canvas id="${{cid}}"></canvas></div>
+            </div>
+        `;
+        
+        setTimeout(() => {{
+            const ctx = document.getElementById(cid);
+            const labels = Object.keys(cat.top_values);
+            const data = Object.values(cat.top_values);
+            
+            new Chart(ctx, {{
+                type: 'doughnut',
+                data: {{
+                    labels: labels,
+                    datasets: [{{
+                        data: data,
+                        backgroundColor: colors,
+                        borderWidth: 0,
+                        hoverOffset: 4
+                    }}]
+                }},
+                options: {{
+                    responsive: true, maintainAspectRatio: false,
+                    plugins: {{ legend: {{ position: 'right', labels: {{ color: '#a1a1aa' }} }} }},
+                    cutout: '70%'
+                }}
+            }});
+        }}, 100);
+    }});
+}}
+
+// 5. Correlation Matrix
+const corrContainer = document.getElementById('corr-matrix');
+if(eda.correlation && eda.correlation.matrix) {{
+    const matrix = eda.correlation.matrix;
+    const cols = Object.keys(matrix);
+    
+    // Header row
+    let headerHtml = `<div class="corr-row"><div class="corr-cell corr-label"></div>`;
+    cols.forEach(c => headerHtml += `<div class="corr-cell" style="color:var(--text-muted); font-size:0.7rem;">${{c.substring(0,6)}}</div>`);
+    headerHtml += `</div>`;
+    corrContainer.innerHTML += headerHtml;
+    
+    // Matrix rows
+    cols.forEach(r => {{
+        let rowHtml = `<div class="corr-row"><div class="corr-cell corr-label">${{r}}</div>`;
+        cols.forEach(c => {{
+            const val = matrix[r][c];
+            let bg = 'rgba(255,255,255,0.02)';
+            if(r !== c) {{
+                const intensity = Math.abs(val);
+                bg = val > 0 ? `rgba(59,130,246,${{intensity}})` : `rgba(239,68,68,${{intensity}})`;
+            }}
+            rowHtml += `<div class="corr-cell" style="background:${{bg}};">${{val.toFixed(2)}}</div>`;
+        }});
+        rowHtml += `</div>`;
+        corrContainer.innerHTML += rowHtml;
+    }});
+}} else {{
+    corrContainer.innerHTML = '<div style="padding:1rem; color:var(--text-muted);">Not enough numeric columns for correlation analysis.</div>';
+}}
+
+</script>
+</body>
+</html>"""
+
+        with open(save_path, "w", encoding="utf-8") as f:
+            f.write(html)
+        print(f"  EDA report saved to: {os.path.abspath(save_path)}")
+        return html
+
+    def run(self, df: pd.DataFrame, save_path: str = "eda_report.html"):
+        print("\n" + "=" * 50)
+        print("  InsightFlow Auto EDA Agent")
+        print("=" * 50)
+
+        print("\nStep 1: Running EDA analyses...")
+        eda_results = run_full_eda(df)
+        ov = eda_results["overview"]
+        print(f"  {ov['rows']} rows, {ov['columns']} columns")
+        print(f"  {len(ov['numeric_cols'])} numeric, {len(ov['categorical_cols'])} categorical")
+        print(f"  {ov['missing_cells']} missing cells, {ov['duplicate_rows']} duplicates")
+        print(f"  {len(eda_results.get('outliers', {}))} columns with outliers")
+
+        print("\nStep 2: Asking AI for insights...")
+        try:
+            ai_insights = self._ask_llm(eda_results)
+            print(f"  Got {len(ai_insights.get('key_findings', []))} findings")
+        except Exception as e:
+            print(f"  AI failed: {e}")
+            ai_insights = {"summary": "AI analysis unavailable.", "key_findings": [], "warnings": [], "recommendations": [], "target_suggestion": "N/A"}
+
+        print("\nStep 3: Generating report...")
+        self._build_html(df, eda_results, ai_insights, save_path)
+
+        print("\n" + "=" * 50)
+        print(f"  Report: {save_path}")
+        print("=" * 50)
+        return eda_results, ai_insights
+
+    def run_file(self, file_path: str, save_path: str = None):
+        from fix import smart_load_file
+        print("\nLoading file...")
+        df = smart_load_file(file_path)
+        if not save_path:
+            base = os.path.splitext(file_path)[0]
+            save_path = f"{base}_eda.html"
+        return self.run(df, save_path=save_path)