See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/342526512

Rain Prediction Using Rule-Based Machine Learning Approach

Article in Advance Sustainable Science Engineering and Technology · May 2020

DOI: 10.26877/asset.v2i1.6019

CITATIONS READS

11 986

4 authors, including:

Muchamad Taufiq Anwar

Universitas Stikubank Semarang
13 PUBLICATIONS 131 CITATIONS

SEE PROFILE

All content following this page was uploaded by Muchamad Taufiq Anwar on 15 February 2021.

The user has requested enhancement of the downloaded file.

Advance Sustainable Science, Engineering and Technology

Rain Prediction Using Rule-Based Machine Learning Approach

Muchamad Taufiq Anwar1, Saptono Nugrohadi2, Vita Tantriyati3, Vikky Aprelia

Windarni4

1
Faculty of Information Technology, Universitas Stikubank, Jl. Tri Lomba Juang No 1
Semarang 50241, Central Java, Indonesia
2
Faculty of Engineering and Informatics, Universitas PGRI Semarang, Jl. Sidodadi-
Timur No.24 Semarang, Central Java 50232, Indonesia
3
Faculty of Information Technology, Universitas Kristen Satya Wacana, Jl. Dr. O.
Notohamidjodjo, Salatiga 50715, Central Java, Indonesia
4
Faculty of Computer Science, Universitas Amikom Yogyakarta, Jl. Ring Road Utara
Yogyakarta, Special Region of Yogyakarta 55283, Indonesia

[email protected]

Abstract. Rain prediction is an important topic that continues to gain attention throughout the
world. The rain has a big impact on various aspects of human life both socially and economically,
for example in agriculture, health, transportation, etc. Rain also affects natural disasters such as
landslides and floods. The various impact of rain on human life prompts us to build a model to
understand and predict rain to provide early warning in various fields/needs such as agriculture,
transportation, etc. This research aims to build a rain prediction model using a rule-based
Machine Learning approach by utilizing historical meteorological data. The experiment using
the J48 method resulted in up to 77.8% accuracy in the training model and gave accurate
prediction results of 86% when tested against actual weather data in 2020.

Keywords: rain prediction, machine learning, J48, data mining

1. Introduction
Rain prediction is an important topic that continues to gain attention throughout the world. The rain has
a big impact on various aspects of human life both socially and economically, for example in agriculture,
health, transportation, etc. Rain also affects natural disasters such as landslides and floods. So much the
impact of rain on human life, then we need a model to understand and predict predictions to provide
early warning in various fields/needs such as agriculture, transportation, etc. Modeling can be made
based on historical weather data that has been recorded by meteorological stations that are scattered in
various locations in Indonesia. This data has been provided by the Climatology, Meteorology, and
Geophysics Agency (BMKG) to be accessed by the public for various purposes including research
purposes. It is known that Machine Learning / Data Mining can be used for weather prediction and
forecasting[1][2]. This study aims to build a rain prediction model using a data mining approach by
0200104-01
Advance Sustainable Science, Engineering and Technology

utilizing historical meteorological data.

2. Methods
2.1. Research on Weather Predictions
Several studies on weather/rain prediction have been conducted. Some studies use a statistical approach
while others use a data mining approach. Research on weather/rain prediction with a data mining /
statistical approach is summarized in Table 1. In weather timeseries research, there are statistical
approaches such as ARIMA, Exponential Smoothing[3], etc and Data Mining / Machine Learning such
as Artificial Neural Networks, etc. [4]. Some studies combine the elements of weather prediction to be
associated with certain phenomena such as Dengue Fever [5], agriculture [6], dan foods[7].

Table 1. Research on weather/rain prediction

Reference Variables Method

[8] Temperature Fuzzy
[9] Barometric pressure, temperature, dew point, humidity, Fuzzy
wind speed
[2] Temperature, rainfall, humidity, exposure time, duration Decision trees, bagging,
of fog, evaporation, wind, atmospheric pressure, number random forests, and
of clouds boosting
[10] Minimum temperature, maximum temperature, rainfall Multiple Linear Regression
[11] Temperature, air pressure, relative humidity, vapor Bayesian
pressure, wind speed
[12] Maximum humidity, average humidity, rainfall Naïve Bayes
[13] Temperature, wind speed, wind direction, humidity, Multiple Linear Regression
atmospheric pressure, rainfall
[14] Maximum temperature, minimum temperature, J48, ANN, dan Naïve Bayes
evaporation, wind speed, cloud cover

2.2. Decision Tree

The Decision Tree (DT) model has a top-down hierarchical structure that describes the rules for dividing
large data sets into small groups given a specific target variable. There are three distinct algorithms for
categorical target variables in the DT model, i.e Entropy Reduction, Gini, and Chi-square tests.
Previously, research on weather forecasting and climate change found that models produced using DT
have small errors compared to other techniques in predicting data mining with large historical data [15],
[16].
The DT model is one of the most powerful and useful for predictions that explore large and complex
data. The mechanism in the DT model is transparent and is produced easily to understand the model for
researchers. Besides, the DT model can convert raw data into information in a simple way, by complying
with a set of rules that can be read by humans. The resulting structure represents a decision or rule for
the classification of datasets. These rules are made to make groups as homogeneous as possible in terms
of response variables. At each step, the input variable is used to divide the observations into groups. If
the specified input values are identified to have a strong relationship with the response values, then all
of these values are grouped in the same branch in the decision tree.
Trees can fit better as the grouping of observation data split into smaller groups (i.e ‘branches’). In
this situation, the DT model will remember data patterns rather than generalize them. The pruning
algorithm in the DT model helps overcome the problem of overfitting by pruning trees using certain
algorithms, namely, CART, CHAID, and C4.5. CART and CHAID both use the Gini Index and Chi-
squared test, respectively, to classify records in the target variable. This research uses the C4.5 / J48
method which uses the measure of Entropy and information gain.

0200104-02
Advance Sustainable Science, Engineering and Technology

2.3. C4.5 / J48 Algorithm

C4.5 is the successor of the Iterative Dichotomiser 3 (ID3) algorithm developed by the same author,
Ross Quinlan, in 1993 [17]. This has several improvements over the original ID3 such as the ability to
handle continuous and discrete attributes and the ability to prune trees after it is created. C4.5 works by
creating trees based on entropy and information gain to select which attributes are useful in classifying
the data. Entropy is a measure of the heterogeneity of data, while information-gain is a measure of how
much information is obtained by comparing entropy before and after separating the dataset based on
certain attributes. Formulas for entropy and information-gain are shown in (1) and (2) respectively.
Pruning in C4.5 is based on the confidence factor. Pruning is useful for minimizing model overfitting
and reducing tree size but in lower model accuracy costs. The well-known implementation of C4.5 is
the J48 function which is written in Java is provided within the Waikato Environment for Knowledge
Analysis (WEKA) software [18]. The pseudocode for the C4.5 algorithm is shown in Algorithm 1 [19].
J48 will produce a tree by which the rules could be easily read by humans. This J48 method has also
been used to find rules for forest fire cases in Indonesia[20]. Research [21] also showed that a decision
tree is very suitable for rain prediction.

c
Entropy ( S ) =  − p log
i =1
i 2 pi (1)


Sv
Gain( S , A) = Entropy ( S ) − Entropy ( S v ) (2)
vValues ( A)
S

Algorithm 1: C4.5
Input: an attribute-valued dataset D
1: Tree = {}
2: if D is ”pure” OR other stopping criteria met then
3: terminate
4: end if
5: for all attribute a ∈ D do
6: Compute information-theoretic criteria if we split a
7: end for
8: a_best=Best attribute according to above-computed criteria
9: Tree = Create a decision node that tests a_best in the root
10: D_v = Induced sub-datasets from D based on a_best
11: for all D_v do
12: Tree v = C4.5(D_v)
13: Attach Tree v to the corresponding branch of Tree
14: end for
15: return Tree

The research methodology is shown in Figure 1. Daily historical weather data was obtained from the
BMKG website for the Tanjung Mas meteorological station, in Semarang City, Indonesia. The original
data consisted of 12 attributes, but for this study, only 8 attributes were used, as shown in Table 1. The
attribute of wind direction was not used since the numerical scale was not appropriate for this study.
One additional attribute is added, i.e the class which shows whether it rained or not on each particular
day. The class is obtained by evaluating the RR (rainfall) attribute column, if RR> 0 then class = ‘rain’;
otherwise, class = ‘norain’. Data cleaning is done to remove entries with missing values. Data is then

0200104-03
Advance Sustainable Science, Engineering and Technology

stored in CSV format and then converted to the ARFF file format to be able being processed using the
WEKA software. Experiments were carried out using the J48 function under the classification tab. The
attributes of the meteorological data are shown in Table 2.

Figure 1. Research Methodology

Table 2. Attributes of the meteorological data

Attribute Data type Description
Tn Numeric Minimum temperature
Tx Numeric Maximum temperature
Tavg Numeric Average temperature
RH_avg Numeric Average Humidity (%)
RR Numeric Rainfall (mm)
ss Numeric Sun exposure time (hours)
ff_x Numeric Maximum wind speed (m/s)
ff_avg Numeric Average wind speed (m/s)

3. Results and Discussion

Meteorological data gathered from the year 2013 to 2019 with a total of 2536 rows of data were used in
this experiment. Evaluation of model accuracy is done by using the 10-fold cross-validation for the
training data and tested against actual weather data in 2020. The training model gives an accuracy of
77.8% whereas the results of experiments against 2020 data gave an accuracy of 86%. The lower
accuracy of the trained model might be caused by the overfitting of the model or that there is a huge
variation in the large amount of training data being used to build the model. These findings also agree
with another research which showed that decision trees and k-mean clustering are best-suited data
mining techniques for weather data, with the increase in the size of the training set, the accuracy is first
increased but then decreased after a certain limit [21]. The model also showed that the factors that predict
rain the most are the average humidity (RH_avg), followed by the minimum temperature (Tn). The hight
accuracy achieved by the J48 method is in line with other research which stated that the Decision Tree
model is better as compared to the other predictive models [14]. The resulted tree which also shows the

0200104-04
Advance Sustainable Science, Engineering and Technology

rules is shown in Figure 2. The model accuracy on various minumum number of cases per leaf is shown
in Table 3.

Figure 2. The (simplified) decision tree produced by J48 (with minumum case of 10 per leaf)

Table 3. The model accuracy on various number of minimum case per leaf

Minimum cases per leaf Model accuracy (%)

2 (default) 76.0
5 76.1
10 77.4
20 77.7
100 77.8

4. Conclusion
A rain prediction model is very useful for human activities. This research attempted to build a rain
prediction model by using a rule-based machine learning approach applied to historical meteorological
data. The decision tree model produced by the J48 algorithm could give an accuracy up to 77.8% from
the training data and give an accuracy of 86% when tested against actual weather data in 2020. The
result showed that rainfall is mainly affected by the average humidity and by minimum temperature for
a particular day of observation. This result gave us a better understanding of the phenomenon of rain
and the model could be used for several purposes such as in agriculture, transportation, etc.

References
[1] M. R. Mahmood, R. K. Patra, R. Raja, and G. R. Sinha, “A novel approach for weather
prediction using forecasting analysis and data mining techniques,” in Innovations in
Electronics and Communication Engineering, Springer, 2019, pp. 479–489.
[2] C. Choi, J. Kim, J. Kim, D. Kim, Y. Bae, and H. S. Kim, “Development of heavy rain damage
prediction model using machine learning based on big data,” Adv. Meteorol., vol. 2018, 2018.
[3] K. D. Hartomo, S. Y. J. Prasetyo, M. T. Anwar, and H. D. Purnomo, “Rainfall Prediction
Model Using Exponential Smoothing Seasonal Planting Index (ESSPI) For Determination of
Crop Planting Pattern,” in Computational Intelligence in the Internet of Things, IGI Global,
2019, pp. 234–255.
[4] N. Mishra, H. K. Soni, S. Sharma, and A. K. Upadhyay, “A comprehensive survey of data
mining techniques on time series data for rainfall prediction,” J. ICT Res. Appl., vol. 11, no. 2,
pp. 168–184, 2017.
[5] N. Agarwal, S. R. Koti, S. Saran, and A. S. Kumar, “Data mining techniques for predicting
dengue outbreak in geospatial domain using weather parameters for New Delhi, India,” Curr.
Sci., vol. 114, no. 11, pp. 2281–2291, 2018.
0200104-05
 Advance Sustainable Science, Engineering and Technology

[6] P. S. Tayde, B. K. Patil, and R. A. Auti, “Applying Data Mining Technique to Predict Annual
Yield of Major Crops,” Int. J., vol. 2, no. 2, 2017.
[7] U. K. Dey, A. H. Masud, and M. N. Uddin, “Rice yield prediction model using data mining,” in
2017 International Conference on Electrical, Computer and Communication Engineering
(ECCE), 2017, pp. 321–326.
[8] K. Kar, N. Thakur, and P. Sanghvi, “Prediction of Rainfall Using Fuzzy Dataset,” 2019.
[9] N. Z. M. Safar, A. A. Ramli, H. Mahdin, D. Ndzi, and K. M. N. K. Khalif, “Rain prediction
using fuzzy rule based system in North-West Malaysia,” Indones. J. Electr. Eng. Comput. Sci.,
vol. 14, no. 3, pp. 1572–1581, 2019.
[10] E. Sreehari and S. Srivastava, “Prediction of Climate Variable using Multiple Linear
Regression,” in 2018 4th International Conference on Computing Communication and
Automation (ICCCA), 2018, pp. 1–4.
[11] V. B. Nikam and B. B. Meshram, “Modeling rainfall prediction using data mining method: A
Bayesian approach,” in 2013 Fifth International Conference on Computational Intelligence,
Modelling and Simulation, 2013, pp. 132–136.
[12] S. Navadia, P. Yadav, J. Thomas, and S. Shaikh, “Weather prediction: a novel approach for
measuring and analyzing weather data,” in 2017 International Conference on I-SMAC (IoT in
Social, Mobile, Analytics and Cloud)(I-SMAC), 2017, pp. 414–417.
[13] N. Anusha, M. S. Chaithanya, and G. J. Reddy, “Weather Prediction Using Multi Linear
Regression Algorithm,” in IOP Conference Series: Materials Science and Engineering, 2019,
vol. 590, no. 1, p. 12034.
[14] N. W. Zamani and S. S. M. Khairi, “A comparative study on data mining techniques for
rainfall prediction in Subang,” in AIP Conference Proceedings, 2018, vol. 2013, no. 1, p.
20042.
[15] A. Joshi, B. Kamble, V. Joshi, K. Kajale, and N. Dhange, “Weather forecasting and climate
changing using data mining application,” Int. J. Adv. Res. Comput. Commun. Eng., vol. 4, no.
3, pp. 19–21, 2015.
[16] F. Sheikh, S. Karthick, D. Malathi, J. S. Sudarsan, and C. Arun, “Analysis of data mining
techniques for weather prediction,” Indian J. Sci. Technol., vol. 9, no. 38, 2016.
[17] J. Quinlan, C4. 5: programs for machine learning. 2014.
[18] M. Hall, E. Frank, G. Holmes, B. Pfahringer, P. Reutemann, and I. H. Witten, “The WEKA
data mining software: an update,” ACM SIGKDD Explor. Newsl., vol. 11, no. 1, pp. 10–18,
2009.
[19] P. Nevlud, M. Bures, L. Kapicak, and J. Zdralek, “Anomaly-based network intrusion detection
methods,” Adv. Electr. Electron. Eng., vol. 11, no. 6, pp. 468–474, 2013.
[20] M. T. Anwar, H. D. Pumomo, S. Y. J. Prasetyo, and K. D. Hartomo, “Decision Tree Learning
Approach To Wildfire Modeling on Peat and Non-Peat Land in Riau Province,” in 2018
International Conference on Advanced Computer Science and Information Systems (ICACSIS),
2018, pp. 409–415.
[21] R. S. Kumar and C. Ramesh, “A study on prediction of rainfall using datamining technique,” in
2016 International Conference on Inventive Computation Technologies (ICICT), 2016, vol. 3,
pp. 1–9.

0200104-06

View publication stats