ARIMA Based Weather Prediction Model using IoT and Open Source Data

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 04 | Apr 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 3268
ARIMA Based Weather Prediction Model using IoT and Open
Source Data
Gunjan Rawat1, Dhruv Patel2, Prajwal Khapare3, Prof Sandhya Deshpande4
1Department of Electronics and Telecommunication Engineering, KJ Somaiya Institute of Engineering and
Information Technology(Autonomous),
(Affiliated to University of Mumbai),Mumbai,India)
2Professor, Department of Electronics and Telecommunication Engineering, KJ Somaiya Institute of Engineering
and Information Technology(Autonomous),
(Affiliated to University of Mumbai),Mumbai,India)
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract -The main objective of our research is to
integrate weather parameters data collected from
IMD(Indian Meteorological Department) and
AccuWeather.com and combine it with our own data
collected from sensors to create a database. This database
is then used to predict and monitor the weather of our
area. The prediction done in this manner will lead to a
more accurate result of the area where the sensors are
located and therefore results in better accuracy and is
more suitable. As it is with respect to an area, it will be
more precise than other Web applications which predict
weather of a whole region compared to a specific area.
This could find it’s application in a variety of fields such as
it can be used by farmers for better prediction, schools,
universities and industries/offices where exist extreme
weather conditions. IMD data is extracted using Python
and a python library created specifically for this purpose
called imdlib. This data constitutes the whole of India so
QGIS can be used for mapping it to a certain area and the
resultant can be stored in the form of a CSV file. The
second part of the database is gathered from sensors
which are connected to a microcontroller which in turn
transfers this data to a wifi module and stores it in the
form of CSV file.The database can then be used to work
with the ARIMA model to predict the data. ARIMA model is
one of the most accurate and easy to implement linear
regression model.
KeyWords: Machine- Learning, IoT, Python ,Arduino,
ARIMA, Linear Regression, Time Series Forcasting
1. INTRODUCTION
Climate change and environmental monitoring have
received much attention recently. With our project we are
proposing a Weather Prediction Model which uses data
from our own sensors(IoT)[1][4] and data available from
IMD(Indian Meteorological Department) as it’s dataset.
IoT-Internet of things is basically a network of physical
objects(sensors) embedded with software and other
technologies. QGIS is a free and open source geographic
information system which allows to analyze and edit
spatial information. Machine learning allows applications
to become more accurate at predicting outcomes. With our
project we are hoping to create a system which will be
able to monitor the weather conditions over controlled
areas like house, industry, agricultural area. Capture
weather parameters like temperature, humidity and rain
using sensors and finally predict the rainfall, temperature
and humidity by implementing Machine learning
Algorithms on the data received from sensors and from
IMD.
1.1 Block Diagram
Figure 1- Sensor Flowchart
Figure 1 is the block diagram which explains the
extraction of data using sensors. The DHT22 sensor senses
the Temperature and humidity while the YL83 is a rain
sensor. This information is then fed to Node MCU .This
information is forwarded further by interfacing the
NodeMCU with its inbuilt ESP8266 Wifi module which
then transfers this data to Thing Speak. The data is then
downloaded in CSV format.[3][4]This is then fed as one of

the inputs for our dataset. The second input will come in
the form of data which is extracted from IMD.
For the second dataset we will be using an inbuilt python
library called imdlib which is specifically used to extract
data from IMD. Using that we will receive the data in CSV
format. Thus the second input for our dataset will be ready
Figure 2- Algorithm Testing Flowchart
After the dataset is ready, the second part comprises
testing the machine learning algorithm. A supervised
machine learning model will be used for this prediction
and after testing a number of different algorithms,
choosing the one with the best accuracy. This will all be
done using Python and its libraries.
2. Methodology
Data will be collected from two streams[2] : sensors and
IMD free available data.
DATA COLLECTION-SENSORS[1][3][4]:
For temperature and humidity we will be using DHT22
sensor, Rain sensor YL 83. These sensors will in turn be
connected to a microcontroller- Arduino UNO. We will use
Arduino IDE for programming this. The arduino will be
connected to a ESP8266 WiFi Module which will transfer
the data to a google sheet/excel as a CSV file using
ThingSpeak. This data will be one of our datasets.
DATA COLLECTION-IMD[2]:
For collecting the IMD data we will be extracting the data
from the .GRD format. An inbuilt python library will be
used to extract the data called ‘imdlib’. This will help us to
get the entire data of a single year . This dataset will then
be mapped to our area using QGIS. This will be our second
dataset.
DATA CLEANING:
All the data has to then be formatted and integrated into
one database.
PREDICTION MODEL:
The final step will be integrating both these datasets and
training our model. We will be using ARIMA time series
forecasting for this which is a type of linear regression
model. This will all be done using Python and its libraries.
3. ARIMA Model
Auto Regressive Integrated Moving Average, abbreviated
as ARIMA, is an Algorithm for forecasting that is centered
on the concept that the data in the previous values of the
time series can alone be utilized in order to predict the
future values.
The primary step is to make the time series stationary in
order to build an ARIMA model. This is because the term
'Auto Regressive' in ARIMA implies a Linear Regression
Model using its lags as predictors. And as we already
know, Linear Regression Models work well for
independent and non-correlated predictors.
In order to make a series stationary, we will utilize the
most common approach that is to subtract the past value
from the present value. Sometimes, depending on the
series complexity, multiple subtractions may be required.
4. Practical Implementation
As the methodology suggests , the flow is followed and the
database is created using the sensor and IMD data. The
connections of the system are as shown in figure 3. The
dht22 is on the breadboard and yl83 independent , both
connected to ESP8266. This is then connected to the
Arduino IDE using a micro USB cable.

Figure 3- Sensor Connections
This sensor data is displayed on ThingSpeak
Figure 4 - ThingSpeak Visualization
After following the methodology and completing the entire
process, the prediction for the maximum temperature for
the month of april is displayed in figure 4. In a similar way
the prediction for the other parameters can be carried out
which are minimum temperature, rainfall and relative
humidity. Jupyter Notebook along with libraries such as
pandas ,numpy, statsmodels, matplotlib were used.
Figure 4 - Maximum temperature prediction
5. Conclusions
Weather of the coming days is correctly predicted(85-90%
accuracy expected) with the help of the entire setup. Data
is successfully collected from the sensors and successfully
extracted from IMD. The resultant datasets are then
formatted, all null/not required values are removed. Both
the datasets are then integrated to form a database which
will serve as input to the algorithm. All this will be
processed in python and using its vast amount of libraries
for machine learning to train the ARIMA model.
6. References
[1] Development of IoT Based Weather Reporting
System”-A F Pauzi1 and M Z Hasan , International
Conference on Technology, Engineering and Sciences
(ICTES) 2020,18 April 2020.

[2] “Smart Weather Forecasting Using Machine
Learning:A Case Study in Tennessee”- A H M Jakaria,Md
Mosharaf Hossain,Mohammad Ashiqur Rahman,arXiv-
Cornell University,25 August 2020.
[3] “Internet of Things (IOT) based Weather
Monitoring System”-Girija C,Harshalatha H,International
Journal of Engineering Research & Technology
(IJERT),2018.
[4] “IOT Based Weather Monitoring and Reporting
System Project”=Anita M. Bhagat, Ashwini G. Thakare,
Kajal A. Molke, Neha S. Muneshwar, Prof. V.
Choudhary,International Journal of Trend in Scientific
Research and Development (IJTSRD),March 2019.

ARIMA Based Weather Prediction Model using IoT and Open Source Data

Recommended

More Related Content

Similar to ARIMA Based Weather Prediction Model using IoT and Open Source Data (20)

More from IRJET Journal (20)

Recently uploaded (20)

ARIMA Based Weather Prediction Model using IoT and Open Source Data