We will study time series analysis, focusing specifically on forecasting sensor data. Sensor data plays a crucial role in various fields, including IoT, environmental monitoring, manufacturing, and more. Accurately predicting future sensor readings is essential for proactive decision-making and optimizing system performance.
- Gain a solid understanding of time series analysis.
- Explore different types of time series data, including continuous and discrete.
- Understand the components of time series, such as trend, seasonality, and irregularity.
- Learn the concept of stationarity and how to test for it using statistical tests like ADF and KPSS.
- Utilize autocorrelation and partial autocorrelation functions (ACF and PACF) to identify dependencies and correlations in sensor data.
- Develop and apply moving average (MA) and autoregressive (AR) models for forecasting sensor data.
- Evaluate model performance using metrics like Root Mean Squared Error (RMSE).
The dataset used in this project consists of IoT sensor data from a chiller. It contains the following columns:
- Time: Timestamp of the sensor reading.
- IOT_Sensor_Reading: Reading obtained from the primary sensor.
- Error_Present: Indicates the presence of an error during the reading.
- Sensor 2: Reading obtained from a subordinate sensor.
- Sensor_Value: The final value to be predicted.
- Language:
Python
- Libraries:
pandas
,numpy
,matplotlib
,scipy.stats
,statsmodels
,seaborn
- Read the data.
- Preprocess the data.
- Perform Exploratory Data Analysis (EDA).
- Check for stationarity in the data.
- Analyze ACF and PACF plots.
- Build the following models:
- Moving Average (MA).
- First-order autoregressive (AR).
- Second/general order autoregressive (AR).
- Third-order autoregressive (AR).
- Fourth-order autoregressive (AR).
- Evaluate the models' performance.
- input: Contains the input data file "Data-Chillers.csv."
- lib: Contains reference materials and a Jupyter Notebook workbook for the project.
- output: Intended to store any project results or output files.
- Readme.md: This file, providing instructions and explanations about the project.
- requirements.txt: Specifies required dependencies.
- src: Contains the source code files, including the main engine file and modules for the machine learning pipeline.