Original article was published by Kahhe Lee on Deep Learning on Medium

Photo Credit: https://unsplash.com/photos/ihMzQV3lleo


Energy Trading Model (Machine Learning — Deep Learning) with AutoCaffe

Over the years, machine learning has proven its effectiveness and efficiencies in reducing companies’ costs. With the vast amount of data generated daily, the “want” of finding the underlying trends in the data, generating consistently accurate forecasts, or to automate processed now becomes a “need”. This resulted in increasing popularity across all industries, hiring data scientists, and implementing machine learning models to keep up with the data.

This Deep Learning Datathon 2020 is organized by Ai4Impact, with SGInnovate as the Deep Tech Partner.

The Team

Our team is called “JJJY”. Together with me, are my teammates: xxx, xxx, xxx. We are all first-year undergraduates from National University of Singapore (NUS) and Singapore Management University (SMU).

Problem Statement:
Build an online financial prediction application that predicts the price of Energy.

In this topic of Energy Trading Model, we will showcase our approach and the step by step to building and incremental testing and fitting of our model to earning a profit! So, let’s dive right in!

The Dataset

There are 2 main datasets: Wind Energy Production and Wind Forecasts.

The source of wind energy production comes from Réseau de transport d’électricité, (RTE) the French energy transmission authority. Denoted as “energy-ile-de-france”.

Wind Forecasts: The wind data come from 2 different wind models, each consists of the forecast from 8 locations in the Ile-de-France region, each forecast consist of 2 variables: wind speed (in m/s) and wind direction as a bearing (degrees North). Thus, 8x2x2= 32 forecasts in all. The data is provided by Terra Weather.

All data provided and generated in the AutoCaffe that we use are from 01 Jan 2017 to the present and is standardized to 1 hourly values.

Data Analysis and Visualization

Here, we use Python 3 — Pandas Libary, Matplotlib, and Seaborn Library to help us conduct data analysis and visualization among the data.

Figure A: Corr Analysis Visualization using Seaborn Heatmap

We first went on to find out the correlations of each forecast against wind energy productions via plotting a Seaborn heatmap of Correlations for all the data/forecast (Refer to Figure A)

Figure B: Corr Analysis against Wind Energy

From the result in Figure B, we notice that there is a clear outlook that the directions of the wind have an extremely low correlation against wind energy. Whereas the speed of the wind, even the lowest forecast have 0.76 correlations against wind energy, highest at 0.81 (“speed-angerville-2-b”).

Figure C: Corr Analysis against the Highest Corr Forecast in Figure B

We then went forward to test all the forecasts against the Highest forecast from the previous result (“speed-angerville-2-b”) as shown in Figure C. This continued to indicate that direction of the wind has low effects on our main objective of predicting wind energy.

Data Analysis — Time Series: Trend

Moving forward to look at the trends in wind energy.