Project 4 : Data Science Frameworks Time Series App

One year Forecast Time Series Models. user inputs a frame work user gets needs to choose method of evaluation ARIMA Make time series data stationary Plot the Correlation and AutoCorrelation Charts Construct ARIMA model based on data(user can choose p,I,q. we will give them best option(if there is time)) Use the model to make predictions(1 year) plot predictions auto ARIMA automate best params Construct ARIMA model based on data Use the model to make predictions(1 year) plot predictions FbProphet Construct model Use the model to make predictions(1 year) plot predictions plot_components() , shows trend and seasonality Link to App

April 18, 2022

Project 3 : Spam detection App

Determine the best Classification model.** We do this by Using the following ML models and hypertunning the models: KNN Find best K value create pipeline and cross validated Fit the model with a range of K values Model prediction Logistic Regression create pipeline and cross validated Model prediction SVM Find initial SVC Then grid search prams(Find best parameters) Pipeline cross validated Fit model with a range of parameters Model Prediction Random Forests Grid search prams(Find best parameters) Pipeline cross validated Fit model with a range of parameters Model Prediction Link to App

April 14, 2022

Project 2 : Programmers Salary Prediction

Determine the salary of a developer. We do this by exploring the following: Since we want to make a salary prediction, we will try two models, Linear Regression model and a lasso regression model Since our data contains 4 no numerical columns, we use OneHotEncoder function from sklean libriary OneHotEncoder converts categorical columns to nemerical columns by getting their dummpy values. I choose OneHotEncoder over get dummies because it can easily be added to my pipeline Next we use a train test split to separate our data into the test set and the training set