Using day to day models — In our previous article, we explored augmenting the feature set by adding technical indicators. In this article, we explore using direct forecasting (also known as day-to-day models) instead of recursive forecasting. The problem statement, feature engineering, training-validation-test split steps remain exactly the same as in the previous article, and will…

Augmenting Features using Technical Indicators — This article repeats the analysis in Part 2, but here we add technical indicators into the list of features for the stock price prediction model. The aim is to observe if technical indicators will improve the performance of the model. Problem Statement Feature Engineering with Technical Indicators Training, Validation, Test split Hyperparameter Tuning Applying…

A Follow-up to Part 1 — So after I published my first article (here) on forecasting stock prices using XGBoost, I received feedback that the signal should first be made stationary before making any predictions, otherwise it is like trying to shoot at a fast-moving and erratic target. …

In 10 (Mostly Easy) Steps — In this article, we will deploy and schedule a Machine Learning model using Amazon EC2 (Elastic Compute Cloud) and Amazon SES (Simple Email Service). This article assumes you have already developed a Machine Learning model using, for example, scikit-learn, similar to what I have done here. If you have not…

A Step-By-Step Walk-Through — There are many machine learning techniques in the wild, but extreme gradient boosting (XGBoost) is one of the most popular. Gradient boosting is a process to convert weak learners to strong learners, in an iterative fashion. The name XGBoost refers to the engineering goal to push the limit of computational…

Exponential smoothing is a family of forecasting methods which computes a weighted average of past observations as the forecast. The weights are decaying exponentially as the observations get older. As a result, the more recent the observation, the higher its weight in the forecast. The family of exponential smoothing methods…

Forecasting is a hard science and requires substantial expertise. For these reasons Facebook open-sourced Prophet, which allows anyone with Python programming skills to implement forecasts on time series easily. You can read more about Prophet on Facebook’s blog or in the paper. In this article, we will experiment with using…

Machine learning has many applications, one of which is to forecast time series. One of the most interesting (or perhaps most profitable) time series to predict are, arguably, stock prices. Recently I read a blog post applying machine learning techniques to stock price prediction. You can read it here. It…