Merge pull request #689 from sanskritiagr/master

Loan Prediction Project

Loan-Prediction-main/Loan-Prediction-main/README.md

@@ -0,0 +1,34 @@
+# Loan-Prediction
+A machine learning model designed to predict whether a loan application will be approved or rejected based on the applicant's personal information and credit history. The model utilizes a dataset containing historical loan application data to train and validate its predictions. 
+## Dataset: 
+The dataset used for training and testing the model consists of various features that include:
+1. Personal information such as gender, marital status, dependents, education, self-employed, and property area.
+2. Financial information such as applicant income, co-applicant income, and credit history.
+3. Loan-specific details such as loan amount, and loan term.
+4. Target variable indicating whether the loan was approved or rejected.
+## Dependencies: 
+Libraries used: Numpy, Pandas, Matplotlib, missingno, seaborn, scipy, sklearn.
+## Visualization:
+To gain insights into the dataset before building the model, I tried to visualize the model using some plots like:</br>
+BarPlot(): I have plotted the crosstab for each class with the loan prediction class(output). </br>
+Histplot(): Showing the distribution of the dataset. </br>
+Heatmap(): Checked if our attributes are codependent on one another. If were so, we could have simply used any one of them but all our attributes are independent.</br>
+Crosstab(): We observed that loan status is not dependent on gender since we have an almost equal ratio for male and female.</br>
+Boxplot(): We observed that we have outliers that need to be taken care of.</br>
+## Data Preprocessing
+Treating NaN values: For the categorical attributes I used 'mode' to fill in the empty cells and for the numerical values, 'mean' was used.</br>
+Treating Outliers: We simply dropped the values that were not in the range. </br>
+SMOTE(): Increasing the data for the minority class.</br>
+MinMaxScaler: Linearly scales them down into a fixed range. </br>
+## Model Training
+1. Linear Regression: Accuracy- 92.59%
+2. KNN: Best Accuracy- 90.74%
+3. SVC: Accuracy- 92.59%
+4. Decision Tree: Best Accuracy- 92.59%
+5. Random Forest: Best accuracy- 92.59%
+6. Gradient Boosting: Best Accuracy- 85.19%
+#### Finally I used MaxVoting to ensemble all the models and got final accuracy of 92.59%.
+
+## After running app.py file
+![image](https://github.com/sanskritiagr/ML-CaPsule/assets/96240350/71ff508e-7277-40b2-82e0-e7854b7c7b34)
+

Loan-Prediction-main/Loan-Prediction-main/app.py

@@ -0,0 +1,53 @@
+# from collections.abc import Buffer
+# from typing_extensions import Buffer
+import numpy as np
+from flask import Flask, request, render_template
+import pickle
+
+#Create an app object using the Flask class. 
+app = Flask(__name__)
+
+#Load the trained model. (Pickle file)
+model = pickle.load(open('models/model.pkl', 'rb'))
+
+#Define the route to be home. 
+#The decorator below links the relative route of the URL to the function it is decorating.
+#Here, home function is with '/', our root directory. 
+#Running the app sends us to index.html.
+#Note that render_template means it looks for the file in the templates folder. 
+
+#use the route() decorator to tell Flask what URL should trigger our function.
+@app.route('/')
+def home():
+    return render_template('index.html')
+
+#You can use the methods argument of the route() decorator to handle different HTTP methods.
+#GET: A GET message is send, and the server returns data
+#POST: Used to send HTML form data to the server.
+#Add Post method to the decorator to allow for form submission. 
+#Redirect to /predict page with the output
+@app.route('/predict',methods=['POST'])
+def predict():
+
+    int_features = [float(x) for x in request.form.values()] #Convert string inputs to float.
+    features = [np.array(int_features)]  #Convert to the form [[a, b]] for input to the model
+    prediction = model.predict(features)  # features Must be in the form [[a, b]]
+    if(prediction[0] == 0):
+        output = "Loan Rejected"
+    else:
+        output = "Loan Approved"
+    # output = round(prediction[0], 2)
+
+    return render_template('index.html', prediction_text='Prediction: {}'.format(output))
+
+
+#When the Python interpreter reads a source file, it first defines a few special variables. 
+#For now, we care about the __name__ variable.
+#If we execute our code in the main program, like in our case here, it assigns
+# __main__ as the name (__name__). 
+#So if we want to run our code right here, we can check if __name__ == __main__
+#if so, execute it here. 
+#If we import this file (module) to another file then __name__ == app (which is the name of this python file).
+
+if __name__ == "__main__":
+    app.run()