Dark Mode
ARIMA Time Series Mumbai Temperature Prediction
Data Science and Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
This product delivers a predictive analysis of temperature for the city of Mumbai, India. The core methodology employs a sophisticated ARIMA time series model, specifically configured as ARIMA (3,0,4). The model exhibits high predictive accuracy, validated at 97.89%, and is considered statistically significant. Key performance indicators include a low Mean Absolute Error (MAE) of 0.587, indicating that the predicted value deviates from the actual value by a minor margin. This focused dataset supports robust time series analysis and facilitates 15-day temperature predictions.
Columns
The resulting data product contains two primary fields derived from the original source file:
- datetime: Represents the date vector, derived from the original file and formatted using the "lubridate" package. This field contains 1781 unique and valid records. The most frequently occurring date is 01-01-2016.
- temp: Denotes temperature measured in degrees centigrade. Across the 1781 valid observations, the mean temperature is 28.3, with a standard deviation of 1.96. The temperature values range between a minimum of 20.2 and a maximum of 32.8.
Distribution
The underlying source is a file named
rainfall.csv, weighing 97.06 kB. The dataset contains 1781 total records, and all data used for modelling was confirmed to be free of missing values. A critical finding for the ARIMA process is the stationarity of the time series, which was verified using the Augmented Dickey-Fuller (ADF) Test. The test resulted in a p-value of 0.01, meeting the criteria for stationarity. Furthermore, diagnostic checks confirm that the residuals are normally distributed in a bell shape, with a constant mean of 0.Usage
The primary application is time series forecasting for temperature within Mumbai. Users can generate forecasts for a 15-day period, specifically ranging from 16th November 2020 to 30th November 2020. The predicted temperatures are provided alongside upper and lower bounds for both 80% and 95% confidence intervals. The product also supports historical model validation, allowing for comparison between the actual and predicted temperatures for a previous 5-day period (11th November 2020 to 15th November 2020). The Root Mean Squared Error (RMSE) of 0.795 further aids in evaluating model suitability.
Coverage
The geographical scope is confined to the city of Mumbai. The final dataset was derived by dropping eight initial columns, focusing only on the two variables required for temperature forecasting. The defined forecast covers the 15-day span from 16th November 2020 to 30th November 2020. The expected update frequency for this product is "Never."
License
CC0: Public Domain
Who Can Use It
- Data Analysts: Individuals requiring validated time series models supported by detailed performance metrics for comparative studies.
- Meteorological Researchers: Those focused on analysing and predicting weather patterns in the Asia region.
- Students and Educators: For practical application of ARIMA methodology, including handling stationary data, interpreting residual diagnostics, and understanding confidence levels.
Dataset Name Suggestions
- Mumbai Temperature ARIMA Forecast (3,0,4)
- ARIMA Time Series Mumbai Temperature Prediction
- High-Accuracy 15-Day Temperature Forecasts for Mumbai
Attributes
Original Data Source: ARIMA Time Series Mumbai Temperature Prediction
QUALITYLoading...