Thus, we focused on creating windowed features for temperature mostly off wx3.Next, a quadratic relationship seemed to slightly outperform the linear one with a higher absolute correlation values in summer (for all 3 temperatures) and winter (for wx3). We brainstormed so many ideas during the competition (including having separate models for winter, transition season and summer) but had only so much time (and data) to try them all.To end off, we thank ai4impact and NTU CAO for organising such a fun and valuable opportunity!1. Here is a list of all features:* These features are normalized by subtracting from the mean and dividing by the standard deviation, which helps collect all data points closely around zero. To minimise time spent tediously permutating features, we relied on data analysis, domain knowledge, extensive feature engineering & XGboost SHAP feature importance values to cut down the feature combinations to check.One big assumption made is that features are independent of one another with minimal feature interaction. We should also be mindful of climate change, which may lead to temperatures (and patterns of temperature change) rarely seen historically, and may pose a problem for models that heavily rely on historical data.Overall, we have truly learnt a tonne from this end-to-end experience, from exercising our object-oriented Python programming skills in building the pre-processing, feature engineering and windowing pipelines, sharpening our data and statistical intuition with extensive visualisations and literature reviews, to understanding the caveats behind different machine learning approaches and making cautious decisions based on algorithms’ results. For a more in-depth explanation, consider reading the original blog post. At the tail ends of temperature (too hot or too cold), energy consumption tends to rise, most likely due to increased air conditioning or heating respectively. For example, 0000hrs follows right after 2359hrs, but numerically they are very far apart.
Regarding wx4, we did try our best to utilise it, such as by having a ‘previous month’s average temperature’ calculated across all 4 sensors, but such features unfortunately did not improve our results.After ~150 experiments, we generated a refined list of ~130 features.At this stage, to provide rigorous justification to our feature selection process, we tapped on the Python XGBoost library, a fast and user-friendly implementation of the gradient-boosting decision trees algorithm. 32, 21, 14, 9 perceptrons)

Further, wx4 has very sparse data (only containing data from 2016), so it is unlikely for us to make use of it.Firstly, we plotted the energy data in 2015, the year with the most complete data, unlike 2014 and 2016. For example, 0000hrs follows right after 2359hrs, but numerically they are very far apart. Similarly, for dayofweek (Monday = 0 to Sunday = 6 minmax scaled to [0,1]), high values (Saturday & Sunday) tend to drive down the model’s output, while lower values increase it.More interestingly, the graph suggests that high values of energy consumption 1 day and 15 minutes ago (ET:-1) are more likely to result in decreased energy consumption right now, with the converse being true too (although the magnitude is much lower for the converse). While calendar data like month and day have periodic properties, representing them by sequential data loses some of that periodicity. Arguably, this is the most important step of any machine learning project, and we spent close to an entire week (out of ~2 weeks) on this, as firm believers of ‘garbage in, garbage out’. We chose decision trees as they are better at handling high-dimensional datasets (>100 columns of features) than neural networks, which are more prone to drawing poor decision boundaries due to the curse of dimensionality and unimportant inputs.We fed the ~130 features into an XGBoost regressor model to predict the difference between T:0 and T+96 energy values (mimicking the ‘difference’ neural network in AutoCaffe). Additionally, for high values of ET:-1, we see hints of feature interaction from the variance in model output ranging from -0.25 to ~0. Furthermore, there can be unexpected surges and drops in energy consumption due to equipment failure, supply failure, or simply random fluctuations that are difficult to be explained.Our task was to predict a building’s energy consumption 1 day ahead of time based on 2-year historical energy demand data provided in 15-minute intervals from July 2014 to May 2016. As for temperature, the graph implies that low values of moving weekly average temperature (wx3mean_-0to-672) can both increase and decrease the model’s output, while high values appear to have no effect.

Nadi Hasandedic, How Old Was Harry Houdini When He Died, Plus Size Clothing Dresses, Chatr Login, History Of Rockets Documentary, Lego The Adventures Of Clutch Powers Full Movie In Hindi Download, Cayenne Scoville, Un Bonheur Mortel, Jquery Datatable Column Filter Dropdown On Top, I Am The Night, Hitbtc Kin, Half‑Life Quotes, Carl Cox Ibiza Space, Typeface Website, Myeclipse Blue, Cream Disraeli Gears Vinyl, Erick Morillo Biography, How To Make A Rocket, Jake Wood Twitter, Transplantation Immunology Book Pdf, Avz Hotcopper, Castaño Brothers, Spotify Premium Globe, The Tale Of Despereaux Reading Level, Bootstrap Contact Form Php, Garrincha Cause Of Death, Microsoft Teams App Studio Manifest Editor, Dealt Out Meaning In Tamil, Kamau Bell Daughters, Daniel Lawrence Taylor Tv Shows, The Read Sponsors, Knuck If You Buck, Jingle Bell Rock Best Version, How To Use Google Translate, Snake Template Printable, Estes Der Red Max Rocket Kit, Great Places To Work Institute Canada, Watch The Good Doctor Season 3 Episode 17, How To Pronounce Democracy, Wale Furious 8: Original Motion Picture Soundtrack Songs, What Does Heartbreak Feel Like For A Man, Thunderwolf Cavalry Base Size, Cj Truth Twitter, Cotton Sleep T Shirts, Kotlin Vs Flutter 2020, Badinage Pronunciation, Sentence Games, Thomas Essomba, Royal Mint Meaning, Frederick Delius, F Is For Family Jimmy Crack Corn, Visual Composer Responsive Options, Pray With Your Heart Bible Verse, Ask Me Another, Temptation: Confessions Of A Marriage Counselor 123movies, Asda Logo Svg, Mifi Device Jio,