Predict Precipitation Using Gain Ratio Biology Essay

Published:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

The population of the world has been increasing substantially. The populous countries like India, seriously lagging behind to provide the basic needs to the people. Food is one of the basic needs that any country has to fulfill. Agriculture is one of the major sectors on which one third of Indian population depends on. The irrigation based countries like India where the water has been the basic resource that forges the plants' growth. The main resource for the irrigation is rainfall which is scientifically a liquid form of precipitation. The atmospheric nimbus clouds are responsible for this precipitation. Prediction of the precipitation is necessary, as it has to be considered during the financial planning of a country. The meteorological departments of every nation are very keen in recording the datasets of precipitation which are huge in content. Hence, data mining is found to be an apt tool which would extract the relation between the datasets and their attributes. A Supervised Learning in Quest is one such data mining algorithm which is eventually a decision tree used to predict the precipitation based on the historical data. The Supervised Learning in Quest decision tree using gain ratio is a statistical analysis for establishing the relation between attribute set and precipitation which furnishes the prediction with an accuracy of 77.78%.

Keywords- Data Mining; Decision Tree; Meteorology; Precipitation; Prediction; Rainfall; SLIQ;

Introduction

The growth of population is one of the major factors that affect adversely the growth of economy of a country. It is essential to ensure adequacy of infrastructure for providing the basic needs of the growing population. The agricultural sector provides the most of the raw materials required for providing products to meet the basic needs. It is obvious that the agricultural productivity depends on water availability wherein precipitation is the primary source of water. The precipitation is due to the thick layers of the clouds in the atmosphere, which would have attained the melting point [26]. The prediction of the precipitation forms a basis for planning economy with improved accuracy. Hence, there is a need to propose the models for improving accuracy in the precipitation prediction.

A mathematical model is an abstract representation of a real-life problem situation. Many mathematical models which represent the real-life problem situations are complex. Hence, solving such complex models involve in performing a large number of arithmetic and logical operations on related data. The invention of the computer improved accuracy and minimized the time in performing those operations. The prediction of precipitation is a complex and uncertain phenomenon that results in the complex mathematical models. The most of the prediction models employ the huge historical data. Here, data mining can be used for predicting the precipitation more accurately.

Data mining tools can be employed in the fields of prediction constitute artificial neural networks, genetic algorithms, ruled based induction, nearest neighbor method, memory based reasoning, linear discriminate analysis and decision trees. The success rate for the prediction of the precipitation by employing different data mining tools reported in the literature is 43.6% [29]. Recently, Prasad et. al proposed to employ Supervised Learning In Quest (SLIQ) decision tree using Gini index for the prediction of the precipitation which resulted in an accuracy of 72.3% [2]. This paper proposes to employ SLIQ decision tree using gain ratio that improves the accuracy from 72.3% to 77.78%.

The rest of the paper is organized as follows: Section II describes relevant work. Section III provides the information about Decision Trees. In section IV, a brief description about the SLIQ Decision tree algorithm is discussed. Section V describes the rules for decision tree. Section VI describes the experimental results. In section VII conclusions are presented and finally in section VIII, the future enhancements are illustrated.

Relevant Work

Research is a continuous process. If anyone imagines that the research on any field is completed and then he/she has to rephrase his/her word of sentence. The research continues beyond this point. In the literature, there are many research findings which are reported for predicting the precipitation with accurate possible rate. Some of them used the traditional methods of the artificial neural networks for the prediction while other methods include the recent developments like Image Processing, Linear Regression and Fuzzy logic and so on.

Frank Silvio Marzano, Giancarlo Rivolta, Erika Coppola, Barbara Tomassetti and Marco Verdecchia used a fully neural network approach to the rainfall field Nowcasting from infrared and microwave passive-sensor imagery aboard [6]. K.Richards and G.D. Sullivan, combined the features of Bayesian scheme for texture analysis of the cloud images which are taken from the ground [7]. C. Jareanpon, W. Pensuwon, R.J. Frank and N. Davey formed radial basis function neural network with a specially designed genetic algorithm [8]. K. Ochiai, H. Suzuki, S. Suzuki, N. Sonehara and Y. Tokunaga stated that the computational time for learning with an acceleration algorithm can be reduced about 10 percent by introducing a pruning algorithm [9]. I.F. Grimes, E. Coppola, M. Verdecchia and G. Visconti presented an approach to cold cloud duration imagery derived from meteosat thermal infrared imagery is used in conjunction with numerical weather model analysis data as an input to an ANN [10]. Thiago N. de Castro, Francisco Souza, Jose M.B. Alves, Ricardo S.T. Pontes, Mosefran B.M. Firmino and Thiago M. de Pereria forecasted seasonal Rainfall using Neo-Fuzzy neuron model [11]. Tuan Zea Tan, Gary Kee Khoon Lee, Shie-Yui Liong, Tian Kuay Lim, Jiawei Chu and Terence Hung IEEE treated the series of rainfall as a continuous time series [12]. Jiansheng Wu Integrated linear regression with ANN. The linear regression extracts linear characteristics of the rainfall [13]. Hui Qi, Ming Zhang and Roderick A. Scofield developed a Multi- Polynomial High Order Neural Network (M-PHONN) [14]. Wint Thida Zaw and Thinn Thu Naing stated that the Multi variables polynomial regression (MPR) is one of the statistical regression methods used to describe the complex nonlinear input and output relationships [15]. C. Kidd and V. Levizzani stated that the rainfall is spatially and temporally highly variable [16]. Sanjay D. Sawaitul, Prof. K.P. Wagh and Dr. P.N. Chatur used the parameters of the weather like wind direction, wind speed, humidity, rainfall and temperature and so on for the classification and prediction of the future weather by using the back propagation algorithm [17]. Soroosh Sorooshian, Kuo-lin Hsu, Bisher Imam and Yang Hong made global precipitation estimation from satellite image by using artificial neural networks [18]. Kesheng Lu and Lingzhi Wang used a bagging sampling technique is used to generate the training sets for combination model based on support vector machine for the rainfall prediction [19]. Grant W. Petty and Witold F. Krajewski discussed in their research methods based on infrared, visible and passive microwave radiation measurements [20].

Decision Tree

Decision tree is an advanced knowledge discovery process with minimum time complexity and has an ease in the implementation. It establishes relationship between the various datasets by discovering the hidden patterns among the datasets which are huge and complex [3, 4], [26, 27]. As it is known fact that, "The only way to get more accuracy is to do more research", which indicates that more and more research has to be done to gain more accurate results. However the research should be carried out by keeping in mind the cost factor. Hence, the scientists have been improving the decision tree algorithms. The use of decision trees have been raised from normal statistical analysis to an effective tool in data mining, text mining, information retrieval and pattern recognition and so on.

The attributes referred in Table I are humidity, temperature, pressure, wind speed and dew point. The amount of water vapor in the air is referred as humidity is invisible in nature. The temperature is the degree of hot or coldness of a body or environment. The temperature is measured in degree centigrade (oC). Atmospheric pressure is the force per unit area exerted against the land surface by the weight of air above the land surface and it is measured in bars. The velocity at which wind is flowing is referred as the wind speed which it is measured in meters per second by an anemometer. Pressure gradient, Rossby waves, jet streams and local weather conditions mainly affect the wind speed which leads to the destruction. Dew point is the temperature at which the air present in the atmosphere can no longer hold all of the water vapor which is mixed with it and some of the water vapor must condense into liquid water.

As it is an established fact that the precipitation generally depends on the various attributes like humidity, temperature, pressure and wind speed and so on. Let us consider a dataset with the similar attributes namely humidity, temperature, pressure, wind Speed and dew point which influence the rainfall and class label as given in Table I. A decision tree is constructed as shown in Fig.1, for the data given in Table 1.

The Table 1 shows 30 days data of humidity, temperature, pressure, wind speed and dew point along with the class label. This is a part of data from Indian Meteorological Department (IMD) for 15 years.

The decision tree is an inverted tree with root node representing the entire dataset which is partitioned into various branches. The leaves of the branches represent class label as shown in Fig.1.

Training Dataset

Day

Humidity

(H)

Temperature

(T)

Pressure

(P)

Wind Speed

(W)

Dew Point

(D)

Class

1

97

24

1005

14

21

Rain

2

85

26

1004

16

21

No Rain

3

91

27

1004

14

21

Rain

4

82

27

1006

16

20

Rain

5

81

26

1007

18

19

No Rain

6

95

26

1007

18

20

Rain

7

95

26

1007

16

20

Rain

8

93

26

1008

18

21

Rain

9

87

24

1005

13

21

Rain

10

88

24

1005

11

21

Rain

11

80

26

1005

14

21

Rain

12

89

26

1005

14

21

Rain

13

86

27

1006

14

21

No Rain

14

86

28

1007

10

22

Rain

15

94

27

1006

14

21

Rain

16

88

26

1004

13

21

No Rain

17

92

27

1005

13

21

Rain

18

86

27

1007

11

21

Rain

19

82

27

1006

11

21

Rain

20

76

27

1007

14

19

No Rain

21

79

27

1008

11

20

No Rain

22

75

27

1008

13

20

No Rain

23

84

27

1007

13

20

No Rain

24

88

26

1006

11

21

Rain

25

86

25

1005

16

19

Rain

26

78

28

1006

13

21

No Rain

27

79

27

1008

13

19

No Rain

28

80

28

1008

8

20

No Rain

29

84

29

1009

6

21

No Rain

30

76

27

1009

6

22

Rain

TABLE II Notations Used in Presenting Sliq Alogrithm

Symbols

Description

D

Set of training tuples with associated class labels

DJ

The set of data tuples in D satisfying outcome J

|D|

The number training tuples in D

C

The class label

Entropy (D)

The information needed to classify a tuple in D

Splitinfo (V)

Normalization to information gain.

Split point

Midpoint of Vi and Vi+1

V

An attribute list

Vi

Set of values in attribute V

Vi+1

Changed Class value in attribute V

Pi

The probability that a tuple in D belongs to class Ci

Di

Values which are greater than or equal to the Split point

Dj

Values which are less than the Split point

The criterion for partitioning dataset at a level is explained in the next section. Decision trees can be used for dataset whether it is continuous or discontinuous. The category of dataset is taken into account which is called as the class label. One of the attributes becomes the root node for the decision tree whereas class label is the leaf node as shown in Fig.1. The knowledge based mining is not so effective in establishing temporal attribute relationships.

SLIQ Decision Tree Algorithm

The decision tree classifier, SLIQ [1] can handle numeric as well as categorical attributes. It employs a pre-sorting technique for reducing the cost of evaluating numeric attributes during the tree-growth phase. Further, the SLIQ using the Minimum Description Length (MDL) principle employs a tree pruning algorithm. It is reported that the SLIQ algorithm is inexpensive in resulting compact and accurate trees [1]. The SLIQ ensures scalability in classifying large datasets consisting of a large number of classes and attributes.

In the construction of the decision tree gain ratio is evaluated at every successive midpoint of the attribute values. However, the efficiency of the SLIQ decision tree algorithm can be improved by evaluating gain ratio only at the midpoints of the attributes where the class information changes. The algorithm for the construction of SLIQ decision tree for the prediction of precipitation is presented below. The notations used are given in Table II.

Overview of SLIQ Decision tree growth and split points

Read dataset into the root node of the SLIQ decision tree

Generate an attribute list for each attribute of the dataset

Sort the attribute lists on attribute value in non-decreasing order

Compute the entropy for the root node

(1)

Compute the Info of attribute list 'V'

(2)

Compute the Gain for each attribute list

Gain (V) = Entropy (D) - Info (V) (3)

Compute split information for a set of values of attribute 'Di' and 'Dj'

Splitinfo (V) = (4)

Determine the Gain Ratio for the attribute values in attribute list 'V'

Gain Ratio (V) = Gain (V) / Splitinfo(V) (5)

Determine maximum gain ratio from among the gain ratios which become the basis for the best split as shown in Table III.

Best Split =Max. Gain Ratio value of attribute (6)

Partition the root node into leaf nodes based on the best split point

Repeat the steps 5 through 10 reading the root node as leaf node until all leaf nodes contain the same class labels.

The primary metric for evaluating the prediction of precipitation is accuracy - the accuracy of a predictor refers to how well a given prediction can give the value of the predicted attribute for new or previously unseen data.

Accuracy = Correct predictions / Total predictions (7)

The ideal goal is to produce compact, accurate trees in a short time with scalability - the SLIQ decision tree algorithm used for the prediction of precipitation takes N input attributes and N class labels as an input and produces the decision tree along with the rules.

The simulated tree shown in Fig. 1 consists of 13 nodes and 7 out of them are depicting rain and the remaining 6 are depicting no rain. The decision tree shown in Fig. 1 NR indicates no rain and R indicate rain.

Fig. 1. Gain Ratio based Decision Tree

TABLE III. Gain Ratio Based Split Value for various attributes

Iteration

Humidity

Temperature

Pressure

Wind Speed

Dew Point

Split Value

Gain Ratio

Split Value

Gain Ratio

Split Value

Gain Ratio

Split Value

Gain Ratio

Split Value

Gain Ratio

Step 1

86.0

0.2791

28.5

0.2149

1007.5

0.1146

9.0

0.0495

20.0

0.1029

Step 2

83.0

0.1602

27.5

0.1602

1007.0

0.2050

17.0

0.0976

20.0

0.1602

Step 3

83.0

0.4459

27.5

0.4459

1006.0

0.205

13.0

0.2367

20.5

0.2367

Step 4

83.0

1.00

27.0

0.3112

1006.0

0.3112

15.0

0.3112

20.5

0.1510

Step 5

77.0

0.2147

27.5

0.0563

1008.5

0.3677

7.0

0.3677

20.5

0.3677

Step 6

83.0

-1.0

27.5

1.0

1007.5

1.0

15.0

1.0

20.5

-1.0

Step 7

88.0

0.0176

25.5

0.0690

1007.0

0.0817

15.0

00690

20.5

0.0579

Step 8

88.0

0.0452

25.5

0.1425

1006.0

0.0452

13.0

0.0859

20.5

0.0631

Step 9

88.0

0.5171

27.0

0.0060

1006.0

0.0060

13.0

0.1284

20.5

-1.0

Step 10

83.0

-1.0

27.0

0.1980

1006.0

0.1188

13.5

0.1908

20.5

-1.0

Step 11

83.0

-1.0

27.0

0.2740

1006.0

0.2740

13.0

0.2740

20.5

-1.0

Step 12

83.0

-1.0

27.0

1.0

1007.5

-1.0

15.0

-1.0

20.5

-1.0

Rules for Decision Tree

Once the decision tree is constructed, there is a possibility that the tree is very large to understand. Hence, to simplify the understanding of the large decision tree the rules are generated.

Rule 1: If [ (humidity < 86.0) and (pressure < 1007.0) and (temperature < 27.5) and (humidity < 83.0)] Then (Prediction = Rain)

Rule 2: If [ (humidity < 86.0) and (pressure < 1007.0) and (temperature < 27.5) and (humidity >= 83.0)] Then (Prediction = NoRain)

Rule 3: If [ (humidity < 86.0) and (pressure < 1007.0) and (temperature >= 27.5)] Then (Prediction = NoRain)

Rule 4: If [ (humidity <86.0) and (pressure >=1007.0) and (dew-point <20.5)] Then (Prediction=NoRain)

Rule 5: If [ (humidity < 86.0) and (pressure >= 1007.0) and (dew-point >= 20.5) and (temperature < 27.5)] Then (Prediction = Rain)

Rule 6: If [ (humidity < 86.0) and (pressure >= 1007.0) and (dew-point >= 20.5) and (temperature >= 27.5)] Then (Prediction = NoRain)

Rule 7: If [ (humidity >= 86.0) and (pressure < 1007.0) and (temperature < 25.5)] Then (Prediction = Rain)

Rule 8: If [ (humidity >= 86.0) and (pressure < 1007.0) and (temperature >= 25.5) and (humidity < 88.0)] Then (Prediction = NoRain)

Rule 9: If [ (humidity >= 86.0) and (pressure < 1007.0) and (temperature >= 25.5) and (humidity >= 88.0) and (wind-speed < 13.5) and (wind-speed < 13.0)] Then (Prediction = Rain)

Rule 10: If [ (humidity >= 86.0) and (pressure < 1007.0) and (temperature >= 25.5) and (humidity >= 88.0) and (wind-speed < 13.5) and (wind-speed >= 13.0) and (temperature < 27.0)] Then (Prediction = NoRain)

Rule 11: If [ (humidity >= 86.0) and (pressure < 1007.0) and (temperature >= 25.5) and (humidity >= 88.0) and (wind-speed < 13.5) and (wind-speed >= 13.0) and (temperature >= 27.0)] Then (Prediction = Rain)

Rule 12: If [ (humidity >= 86.0) and (pressure < 1007.0) and (temperature >= 25.5) and (humidity >= 88.0) and (wind-speed >= 13.5)] Then (Prediction=Rain)

Rule 13: If [ (humidity >= 86.0) and (pressure >= 1007.0)] Then (Prediction = Rain)

Experimental Results

The data taken for the training needs to be sorted during the initial stage of the tree growth phase of decision tree construction [3]. As per the training data, humidity is the first attribute. Take the humidity attribute and its corresponding class label as a pair, identify the split points whenever there is a change in the class label. The better split point needs to be found for increasing the accuracy of prediction. For every split point identified find the midpoint for the changed class labels and proceed until it reaches the end of the data as shown in Table IV.

From the Table IV it is clearly visible that there is a change in the class label for the first time at the 3rd position. Mark it as split point and take the midpoint value of 2nd and 3rd class label values i.e. midpoint (76, 76) =76. Similarly the second split point occurs at 4th position. Mark it as split point and take the midpoint value of 3rd and 4th class label values i.e. midpoint (76, 78) = 77. Proceeding in this order there are nine split points as the class label is changing at nine positions.

Repeat the procedure to find out the split points for the attribute temperature shown in Table V, attribute pressure shown in Table VI, attribute wind speed shown in Table VII and attribute dew point shown in Table VIII.

TABLE IV. Dataset sorting on humidity

Humidity

Split PointClass

75

No Rain

76

76

77

80.0

80.5

81.5

83.0

86.0

87.5

88.0No Rain

76

Rain

78

No Rain

79

No Rain

79

No Rain

80

No Rain

80

Rain

81

No Rain

82

Rain

82

Rain

84

No Rain

84

No Rain

85

No Rain

86

No Rain

86

Rain

86

Rain

86

Rain

87

Rain

88

No Rain

88

Rain

88

Rain

89

Rain

91

Rain

92

Rain

93

Rain

94

Rain

95

Rain

95

Rain

97

Rain

TABLE V. Dataset sorting on temperature

Temperature

Split PointClass

24

Rain

24

Rain

24

25.5

26

26.5

27

27.5

28.0

28.5Rain

25

Rain

26

No Rain

26

No Rain

26

No Rain

26

Rain

26

Rain

26

Rain

26

Rain

26

Rain

26

Rain

27

No Rain

27

No Rain

27

No Rain

27

No Rain

27

No Rain

27

No Rain

27

Rain

27

Rain

27

Rain

27

Rain

27

Rain

27

Rain

27

Rain

28

No Rain

28

No Rain

28

Rain

29

No Rain

TABLE VI. Dataset sorting on pressure

Pressure

Split PointClass

1004

No Rain

1004

1004

1005.5

1006

1006.5

1007

1007.5

1008

1008.5

1009No Rain

1004

Rain

1005

Rain

1005

Rain

1005

Rain

1005

Rain

1005

Rain

1005

Rain

1005

Rain

1006

No Rain

1006

No Rain

1006

Rain

1006

Rain

1006

Rain

1006

Rain

1007

No Rain

1007

No Rain

1007

No Rain

1007

Rain

1007

Rain

1007

Rain

1007

Rain

1008

No Rain

1008

No Rain

1008

No Rain

1008

No Rain

1008

Rain

1009

No Rain

1009

Rain

TABLE VII. Dataset sorting on wind speed

Wind Speed

Split PointClass

6

6

7

9

10.5

11

12

13

13.5

14

15

16

17

18

No Rain

6

Rain

8

No Rain

10

Rain

11

No Rain

11

Rain

11

Rain

11

Rain

11

Rain

13

No Rain

13

No Rain

13

No Rain

13

No Rain

13

No Rain

13

Rain

13

Rain

14

No Rain

14

No Rain

14

Rain

14

Rain

14

Rain

14

Rain

14

Rain

16

No Rain

16

Rain

16

Rain

16

Rain

18

No Rain

18

Rain

18

Rain

TABLE VIII. Dataset sorting on dew point

Dew Point

Split PointClass

19

No Rain

19

No Rain

19

19

19.5

20

20.5

21No Rain

19

Rain

20

No Rain

20

No Rain

20

No Rain

20

No Rain

20

Rain

20

Rain

20

Rain

21

No Rain

21

No Rain

21

No Rain

21

No Rain

21

No Rain

21

Rain

21

Rain

21

Rain

21

Rain

21

Rain

21

Rain

21

Rain

21

Rain

21

Rain

21

Rain

21

Rain

21

Rain

22

Rain

22

Rain

Now, compare all the split points' gain ratio values and the value which is maximum is the best split point for that attribute as shown in Table III. The gain value obtained for the attribute is to be divided by split info value of class label, in order to obtain the gain ratio value for that attribute and is shown in equation (9).

Gain Ratio (V) = Gain (V) / Split info (V) (9)

Repeat the above procedure by taking the temperature attribute along with the class label, Pressure attribute along with the class label, wind speed attribute along with the class label and finally dew point attribute along with the class label to get the best split points. Choose the maximum gain ratio value and that itself becomes the root node. Based on the threshold value of the root node generates the tree. Repeat the procedure till it is terminated with a unique class label.

The gain ratio is generally used to measure the inequalities among the statistical data and its frequencies. So far, its use is limited for the analysis of wealth and income of the economic countries. Due to the inequalities present in the probabilities, there may be some error. But, irrespective of its limitation present, it has a wide variety of the applications in statistical analysis.

The gain ratio is used here for the construction of the decision tree where the roots and sub-roots are classified. The use of the gain ratio for the rainfall analysis is quite apt because of the irregularities present in the statistical data of the precipitation. The precipitation data is used does not follow an order in other words a sequential path. This may be due to the inequalities of the present attribute with former attribute. This may change to a great extent or to some extent depending on the Mother Nature.

Some experiments have been conducted on real data to analyze the accuracy of the tree. We have used the dataset from the accuweather.com of Indian Meteorological Department. The goal is to predict the precipitation for rainfall. The dataset consists of 15 years of data from the year 1998 to 2012 containing of 5230 examples. Out of 15 years data 9 years data is used as training dataset and the remaining 6 years data is used as test dataset.

It has been found in Table IX, the distinction between the success rate of prediction and time. It can also be observed, that the maximum efficiency obtained is 74.1% on one year dataset, 77.47% for two years dataset, 77.38% for three years dataset, 77.17% for four years dataset, 77.39% for a five 5 years dataset and 77.78 % for a six years dataset. The average efficiency has been found to be 77.78%. Though, this contributes a decent efficiency or success rate, the other methods of back propagation neural networks [7,8], [12-15], linear discriminate statistical analysis [16] and J48 are analyzed to select the best performing method of prediction of the precipitation.

The published results for this dataset are: 64.3% accuracy for backpropagation, 58% for a linear discriminant and 68.6% for J48. Using the same training and test datasets, Since the average accuracy using SLIQ with gain ratio is 77.78% as shown in Fig. 4, SLIQ using gain ratio can be considered as the best performing method for the prediction of precipitation.

TABLE IX. Result showing the Accuracy and Time of Response

No. of records

Correct predictions

In correct predictions

Accuracy (%)

Time

(Sec)

363

269

94

74.104

37

728

566

162

77.471

40

995

770

225

76.381

42

1262

974

288

77.175

44

1619

1253

366

77.392

46

1981

1541

440

77.778

47

Fig 2. No. of Records Vs Correct Predictions

Fig 3. No. of Records Vs incorrect Predictions

Fig 4. No. of Records Vs Accuracy (%)

Fig 5. No. of Records Vs Time (Sec)

The variation of correct predictions with dataset is shown in Fig. 2. This indicates that there lies a linear relationship between correct predictions and number of records in the dataset.

The variation of incorrect predictions with dataset is shown in Fig.3. This indicates that there lies a non linear relationship between incorrect predictions and number of records in the dataset. From the above graph, the number of incorrect predictions follows a decreasing trend up to 600 records and thereafter increases non linearly. The variation of accuracy with the dataset is plotted in Fig. 4. The variation of time of response towards dataset is plotted in Fig. 5.

Conclusion

The economy of a nation depends on agricultural productivity which is the basis for formulating economic policy. The agricultural productivity depends on the availability of water. The precipitation is the major source of water which depends on various attributes like humidity, pressure, temperature, wind speed, dew point and so on. Hence, the prediction of precipitation becomes a difficult task as it has to consider many parameters. Many techniques such as neural networks, artificial intelligence, used for prediction of precipitation have less accuracy. So far, the maximum accuracy reported is 72.3%. This study employed SLIQ decision tree using gain ratio as splitting criterion. For evaluating the effectiveness of this model the historical data obtained from IMD is applied. It is found that the method proposed in this paper gives higher accuracy when compared to the other models.

Future Enhancements

In this paper, we highlighted gain ratio based SLIQ decision tree algorithm, which gives maximum accuracy. For future implementation various other decision tree algorithms like CART, SPRINT, ELEGANT, EC4.5 with additional parameters can be developed. A decision tree must be developed for the dynamic mode of data rather than static mode.

Writing Services

Essay Writing
Service

Find out how the very best essay writing service can help you accomplish more and achieve higher marks today.

Assignment Writing Service

From complicated assignments to tricky tasks, our experts can tackle virtually any question thrown at them.

Dissertation Writing Service

A dissertation (also known as a thesis or research project) is probably the most important piece of work for any student! From full dissertations to individual chapters, we’re on hand to support you.

Coursework Writing Service

Our expert qualified writers can help you get your coursework right first time, every time.

Dissertation Proposal Service

The first step to completing a dissertation is to create a proposal that talks about what you wish to do. Our experts can design suitable methodologies - perfect to help you get started with a dissertation.

Report Writing
Service

Reports for any audience. Perfectly structured, professionally written, and tailored to suit your exact requirements.

Essay Skeleton Answer Service

If you’re just looking for some help to get started on an essay, our outline service provides you with a perfect essay plan.

Marking & Proofreading Service

Not sure if your work is hitting the mark? Struggling to get feedback from your lecturer? Our premium marking service was created just for you - get the feedback you deserve now.

Exam Revision
Service

Exams can be one of the most stressful experiences you’ll ever have! Revision is key, and we’re here to help. With custom created revision notes and exam answers, you’ll never feel underprepared again.