ML EDA and Prediction Flask App with Gemini API

This project is a Flask-based web application that allows you to:

• Upload a CSV dataset
• Perform Exploratory Data Analysis (EDA)
• Train machine learning models (Regression or Classification) with automated feature selection using the Gemini API
• Handle missing values by dropping features with >50% missing values and imputing others
• Make predictions using the trained model via a web interface

---

Features

• CSV File Upload:
  Upload your dataset in CSV format through a simple web interface.

• Exploratory Data Analysis (EDA):
  Generate a comprehensive EDA report with:

  • Dataset overview (shape, first few rows, basic statistics)
  • Missing values heatmap
  • Histograms with KDE overlays for numeric features
  • Boxplots for numeric features
  • Correlation heatmap for numeric features
  • Count plots for categorical features
  • Pairplot for feature relationships (if applicable)

• Feature Selection with Gemini API:
  Automatically select the most relevant features for the model. The application sanitizes column names (e.g., renaming "Sex" to "Gender_Encoded") to bypass the Gemini API safety filters.

• Missing Value Handling:
  For each feature:

  • Drop the feature if more than 50% of the values are missing.
  • Fill missing numeric values with the column mean.
  • Fill missing categorical values with the column mode.

• Model Training & Prediction:
  Train a model (Linear Regression for regression tasks or RandomForestClassifier for classification tasks) and make predictions based on user input.

• User-Friendly Interface:

  • The Perform EDA link opens the analysis report in a new tab.
  • Dynamic forms are used for loading features and making predictions.

---

Prerequisites

• Python 3.x
• Flask
• Pandas
• Scikit-Learn
• Requests
• Matplotlib
• Seaborn

You can install the required Python packages using pip:

pip install -r requirements.txt

---

Gemini API Key

Before running the application, you must add your Gemini API key:

1. Open the gemini_api.py file.

2. Replace the placeholder key with your actual Gemini API key:

   GEMINI_API_KEY = "YOUR_ACTUAL_GEMINI_API_KEY"

Get your API key from Google AI Studio.

---

How to Run the App

Start the Flask application by running:

python app.py

Then, open your browser and navigate to: http://127.0.0.1:5000/

---

Usage

Training a Model

• Upload a CSV file:
  Use the homepage to upload your dataset.

• Select the target column:
  Specify which column you want to predict.

• Choose Classification or Regression:
  Select the appropriate model type.

• Click Train Model:
  The app uses Gemini to automatically select the best features and trains the model.

Making Predictions

• Select a trained model:
  Provide the path to your saved model.

• Click Load Features:
  The app will auto-fill input fields based on the model's features.

• Enter values & Click Predict:
  Input your values and view the prediction result.

---

Troubleshooting

• Gemini API Blocked?
  The app automatically renames "Sex" to "Gender_Encoded" to avoid triggering safety filters.

• Gemini Fails?
  The app falls back to manual feature selection if the API fails to return valid features.

---

Contributors

• Ritkaar Singh - Developer & Maintainer
  Open to collaboration!