Time Series Forecasting

About time series prediction using SmallTrain

SmallTrain supports various data. This section describes a time series forecasting project using SmallTrain.

AI is being used in various fields such as time-series data prediction, image recognition, and voice recognition.
Among them, Geek Guild has been working on forecasting using time series data from an early stage. With SmallTrain, you can get highly accurate results even with small data.

We have experienced highly accurate forecasts using the following time series data.

  • Hotel room price forecast (dynamic pricing)
    …We are working on not only time series prediction using Convolutional Newral network (CNN) but also reinforcement learning.

  • Power Generation Forecasting …In 2018, the best mean square error (RMSE) reached 0.5%. Highly accurate forecasts have been achieved even with minimum layers of CNNs.

  • Power Consumption Forecasting …In the power consumption forecasting, the previous study of power consumption prediction by CNN has a record of a mean square error of RMSE0.677, but SmallTrain has achieved the world’s highest accuracy of RMSE0.667 (based on 2018 data)

  • Power Plant Anomaly Detection …Start a project using SmallTrain

  • Others …Various Small Train projects such as factory abnormality detection

Anomaly Detection of Manufacturing

As an example of time series prediction using SmallTrain, explain the anomaly detection in manufacturing system.

Dynamic Pricing in Hotels

SmallTrain AI model for dynamic pricing that predicts hotel room prices

Power Generation

Power generation/power consumption forecasting and power plant anomaly detection