SVM_stock_price_predictor

Stock price prediction program that uses different SVM(Support Vector Model) models (specifically RBF, Linear, and Polynomial regression models) to accurately predict the future stock price of a stock from past data.

Process:

Retrieve Data for Training

• Retrieve past stock price of a stock in a given time frame (I used 28 days) and collect that data into a csv file. I used Yahoo Finance for this, as you can set the timeframe you like and download a csv file straight from there.

Reshape the data

• use Numpy to reshape the array into a 2D array that we can then create regression models

Create and train our three models (Linear, Polynomial, and RBF Regression)

• I used the SVR method found in scikit.learn and then fit the models into our datasets

Plot results

• first generate a scatter plot of our data points
• plot our three regression models
• label our axes

Get our predictions

• using the three helper functions I created (get_data, train_models, plot_results) we can then process our data to get predictions using our final function 'process_data'

• we get our dates and prices array using get_data

• we train our models

• we plot our results

• we then reshape our date array into a 2D array (one row of samples and one column of features)

• we then use '.predict' to get our individual predictions, dependent on type of regression, using our reshaped input data

• we then return these predictions