Each project comes with 2-5 hours of micro-videos explaining the solution.
Code & Dataset
Get access to 50+ solved projects with iPython notebooks and datasets.
Add project experience to your Linkedin/Github profiles.
Understanding the problem statement
Importing the necessary libraries and understanding its use
Importing the dataset directly from github
Performing basic EDA and checking for the null values
Filling the null values using appropriate methods
Finding median, average and merging the data
Feature engineering with the date
Plotting time-series graphs for visualization
Drawing a heatmap with the numeric values using Seaborn
Finding lag and lead of a time series
Using groupby function for combined analysis of variables
Differentiating a time series
Performing train_test_split to divide the dataset into train and test
Using r2_score and mean_absolute_error as evaluation metrics
Using Adaboost Regressor for making predictions
Applying the ARIMA time series model for training and making predictions
Applying Facebook Prophet model for making predictions
Visualizing the result using graphs
Selecting the best model and making the final predictions
There are various methods to perform time series forecasting. Traditionally people have used AR, MA or ARIMA based models to perform forecasting. Prophet is an open source forecasting tool built by Facebook. It can be used for time series modeling and forecasting trends in the future. The advantage of using Prophet over traditional libraries is that one does not need to know the technicalities of time series, domain knowledge is not really required to do time series forecasting. In this Hackerday we are going to use Prophet vs other methods to do the benchmarking.