A comparison of a performance of various machine learning models to predict the sales components is presented in this paper. The general aim of this thesis was to find a suitable machine learning model that can fit and well forecast the sales components before they happen and thus create the purchase orders before the product runs out of stock or sold. The dataset used in the thesis is from a product supplier consisting of product sales data for about a thousand assorted products on a time of three years. Firstly, a Literature review used to find suitable machine learning algorithms and then based on the results obtained, an experiment was performed to evaluate the performances of machine learning algorithms. Results from the literature review shown that regression algorithms namely Supports Vector Machine Regression, Ridge Regression, Gradient Boosting Regression, and Random Forest Regression are suitable algorithms and results from the experiment showed that gradient boosting has performed well than the other machine learning algorithms for the chosen dataset, The range of supervised and unsupervised algorithms is provided. The different models were compared using the performance metrics such as mean squared error (MSE), Mean Absolute Error (MAE) and Root Squared (R^2) called coefficient of determination as well based on the dataset’s values against the predicted values. The results shows that the Gradient boosting have shown the highly fitting capability and as well with the highly accuracy compared to other Machine learning models. To conclude After the experimentation and the analysis, the Gradient boosting algorithm has been performed well when compared with the performances of the other algorithms and therefore, gradient boosting is chosen as the optimal algorithm for performing the sales forecasting of the sales components at Maison Gil Lt.

Keywords : Time Series Forecasting, Sales Forecasting, Mean Absolute Error