Water Quality Index Calculation of River Ganga using Decision Tree Algorithm

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 3911
Water Quality Index Calculation of River Ganga using
Decision Tree Algorithm
Kaushiki Agrahari1, Ankur Kashyap2, Sakshi3 , Shashank Kumar4 , Shivansh Sharma5
Shyam Dwivedi6
1,2,3,4,5Student of B. Tech Final year, Department of Computer Science and Engineering,
Rameshwaram Institute of Technology & Management, Affiliated to AKTU, Lucknow, U.P ,India
6Assistant Professor & Head of Department, Department of Computer Science and Engineering,
Rameshwaram Institute of Technology & Management, Affiliated to AKTU, Lucknow, U.P, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The Ganga is not only a pious river, but also a way of life for many citizens of India .It has cultural, spiritualaswellas
scientific importance. Our project research is based entirely on Machine Learning Application . Rapid urban sprawl, industrial
development and high demand for water have caused major problems with water quality degradation and deterioration .
Therefore, the aim of our project is to analyze the water quality of the Ganga River for three different seasons basically Summer,
Monsoon and Winter to assess whether Ganga river water is potable or not. Rivers are under heavy degradation due to human
activities such as dumping of waste, industrial activities, mining in the river, the water level of the river has dropped dramatically
which has affected marine life and human health. The method we use in our research is to design a machine learning modelbased
on ML algorithm which will calculate WQI and provide measurement of river waterqualitythatwill beusedtodetermine whether
the water is drinkable or not. It will be a python based WQI calculator.
Key Words: Machine Learning, Decision Tree, WQI
1. INTRODUCTION
Ganga River is considered as the most prominent river of India .Using latest technologies like Machine Learning and Python
Framework Flask , we are going to create a machine learning model which will predict the quality of the Ganga river water
where the measuring parameters dataset is collected from the website of UCI Repository .Based on this collected data from
different cities where Ganga riverbed is huge, the model will be trained and the prediction algorithm which will be most
accurate and precise will be used.
1.1 Machine Learning (ML) is the study of computer science where algorithms can be used to make decisions on its own
and decisions can be improved through learning and past experiences. It is one of the subsets of artificial intelligence.Machine
learning algorithms build models based on training datasets or sample data to make predictions or decisions without being
explicitly programmed. Now a days , Machine Learning can be seen in every field such as the health sector, finance sector,
Aeronautical sector ,Agriculture ,email filtering, voice recognition, speech recognition, and computer vision, etc and it is quite
not easy to develop traditional algorithms to perform the tasks which ML models can easily do . Machine learning uses data
and learns from that data for better future predictions. Machine learning models canbe improvisedmorebyeffectivehandling
of training dataset so that no overfitting and underfitting comes into picture.
The three main types of machine learning tasks: supervised learning, unsupervised learning, and reinforcement learning.

1.1.1 Supervised Learning, a set of examples (i.e. the training set) is submitted to the system as input during the training
phase. Each input is labeled with the desired output value, so the system knows what the output will look like when the input
comes in. For example, consider some experimental observations that can be grouped into N differentcategories.Sowehavea
training set (a pair of sequences) {(x1,y1),(x2,y2)…..(xn,yn)}, where xi is the input and yi is the class of the corresponding
output. Training is performed by minimizing a specificcostfunction representingthe bindingsfromtheinputxiandthedesired
output yi.
1.1.2 Unsupervised Learning provides training examples that are not labeled with the class to whichtheybelong.Therefore,
the system develops and organizes data, looking for common characteristics among them and making changes based on
internal knowledge.
1.1.3 Reinforcement Learning is a machine learning training method based on rewarding desirable behaviors and or
punishing unwanted behaviors. In general, reinforcement learning agents can recognize, interpret, take action, and learn
through trial and error in their environment.
1.2 Machine Learning Algorithm (Decision Tree )
Decision Tree is a non-parametric supervised learning machine learning algorithmusedforclassificationandregression.Itisa
tree structured classifier where internal nodes represent the features of the dataset, branches represent the decisionruleand
leaf nodes represent the outcome and do not contain further branches.Theobjectiveis tocreatea model thatpredictsthevalue
of a target variable by learning a training dataset of decision rules taken from the data features. Decision Tree Classifier is a
classification algorithm capable of performing multiple class classification on a dataset.Aswithotherclassifiers,DecisionTree
Classifier takes as input two arrays: an array X, sparse or dense, of size n samples, n-features holding the trainingsamples,and
an array Y of integer values, size n-samples, holding the class labels for the training samples.
A decision tree is a one of the types of supervised machine learning algorithms where both input and output data are labeled.
It can be used for both Categorical and Continuous Output . The dataset is continuously divided into sub trees based on some
parameter and calculation measure until the final output is achieved. A Decision Tree can be explained by two entities,
decision nodes and leaf nodes . Leaf nodes are decisions or end results where further division of nodes will not be possible
.Decision nodes are the nodes where data is split and further sub -trees are generated.
Decision Tree Main Components :
 Root Node : The node from where the decision tree starts .It represents the entire dataset which further gets divided
into two or more sets.
 Leaf Node: The leaf nodes are the final outputs and tree can not be segregated furthermore after getting the leaf node
as it is the final output .
 Pruning : The process of removing unwanted branches from the tree to reduce complexity and ambiguity .
 Child Node: The segregated nodes are child nodes.

1.2.1 Algorithm for Decision Tree
Step 1 : Begin the tree with the root node which contains the complete dataset.
Step 2: The best attribute in the dataset using Attribute Selection Measure is selected.
Step 3: Divide the dataset into subsets that contain all possible values to achieve the best attribute.
Step 4 : The decision-tree-node which contains the best attribute is generated.
Step 5 : Recursively new decision trees are created using the subsets of the datasetcreatedinstep3.Thisprocessiscontinued
until the final stage is reached called leaf node when further segregation will not be possible.
There are two types of decision trees:-
Classification trees : The decision variable is 0 or 1 type or Yes/No called as Categorical Value.
Regression trees : The decision or the outcome variable is Continuous in nature , for example, a number like 187.
There are many algorithms by which Decision Trees can be constructed , but one of the best is called the ID3 Algorithm. ID3
Stands for Iterative Dichotomiser .
1.2.2 Entropy: Entropy, also known as Shannon Entropy is denoted by H(S) for a finite set S, which is the measure of the
amount of uncertainty or randomness in the data. In the Machine Learning paradigm, entropy measures unpredictability and
impurity, it is related to randomness in the information being processed in the machine learning project.
Entropy(S) = -(P(y) P(y)-P(n) P(n))
1.2.3 Information gain: Information gain is also called the Kullback-Leibler divergence, denoted IG(S,A) for a set S, which is
the effective entropy change after selecting a certain attribute A. It measures the relative entropy change with respect to
independent variables.
Information Gain= Entropy(S)-{(weighted average )* Entropy(Each feature)}
Alternatively ,
where IG(S, A) is the Information gain by implementing feature A, H(S) is the Entropy of the entire dataset,
is the Entropy after applying the feature A and P(x) is the probability of event x.
1.3. Iterative Dichotomiser ID3 Algorithm is a classification algorithm that follows the greedy approach of decision
trees by selecting the optimal attributes that produce the maximum information gain (IG) or minimum entropy (H).
I(p,n) = entropy of a dataset
. = weighted summation of the logs of the probabilities of the each possible outcome
=-[(p/(p+n)) (p/(p+n)) + (n/(p+n)) (n/(p+n))]
ID3 procedure :
 Calculate the entropy of the dataset.
 For each attribute/feature.
 Calculate entropy for all its categorical values.

 Calculate information gain of the feature/attribute.
 Find the feature with maximum information gain.
 Choose the feature with maximum information gain as the root node for the tree.
 Repeat the above steps until the desired tree is obtained.
2. PROJECT OBJECTIVES
The objective of the machine learning project is to create a Water Quality Index (WQI) Calculator using python and Flask
which will analyze the quality of Ganga river water and decide whether the water is safe for drinking by scaling the water in
the score of (0-100).
3. IMPLEMENTATION & METHODOLOGY
3.1 Study Area: The present study was performed to study the surface water quality in the riverbed movingarea ofGanga
River at district namely; Varanasi, Kolkata, Jahangirpur, and Tribeni.
3.2 Sampling Procedure and Methods of Analysis: The physicochemical parameters used in this study are pH,
Temperature, Total Dissolved Solid, Biological Oxygen Demand (BOD), Dissolved Oxygen , Turbidity, Nitrate, Phosphate and
Fecal Coliform. We have combined data set of our study sites in three different seasons namely Summer, MonsoonandWinter
for each of the districts. We have trained the combined data-set on the basis of machine learning algorithms and used Water
Quality Index Calculation for Quality checking of water for different cities in different seasons and to check water is drinkable
or not. The parameters which are finalized after pre-processing of the datasetaresomehowinterdependentoneachother. The
model will be trained on the basis of these nine physicochemical parameters whichwill befurtherusedforcalculationof Water
Quality Index and deciding water is drinkable or not.
Flow of Model Creation
Table 1 : Table depicting mean values of parameters in Study Site

3.3 Graphical Representation of Water Quality Index Calculating Parameters
These graphical plotting of the nine parameters defines how the attributes are fluctuating in different seasons
3.3.1 pH: pH is a scaling methodology which is used to determine theacidityorbasicityofanaqueoussolution.LowpHvalues
denotes acidic character, while higher values denotes basic character of the solution. The pH scale is (0-14),where (0-7)
denotes acidic character and (7-14) denotes basic character of the solution.The basic pH range for surface water is: (6.5 - 8.5)
.pH is the crucial parameter associated with the other calculating parameters for WQI calculation. As pHcanbe easily affected
by other chemicals in water, pH is necessary to be accurate for best results of the model. pH iscalculatedinlogarithmicunitsof
base 10 which means each number represents a 10-fold change in acidity or basicity depending on the pH value.
(pH = − (H +).
3.3.2 Temperature : The surface water temperature is usually between (0-30) degrees Celsius.Thetemperatureisthewater
parameter which refers to how cold or warm the water is. In monitoring water quality and in field of science, temperature is
basically measured in degrees Celsius. Water temperature affects almost every other water quality parameter. Temperature
affects other water quality parameters to a great extent ,also can change the physical and chemical properties of water too.
The varied seasonal temperature fluctuations may be due to changes in air temperature, solar angle, weather events, global
warming and the amount of material associated with the stream and water features.

3.3.3 Total Dissolved Solid : TDS is a measure of the dissolved solvents found in an aqueous solution. In fact, this isanything
that contaminates the purity of water. Some of these solvents can cause a large number of health problems.TDS is an indicator
of water quality. Having a high TDS does not always mean that water is dangerous. For example, some mineral water has very
high levels of solids inside and this does not seem to be dangerous, but that is because real solid materials are. This is usually
Ca, Mg or other substances that may not be harmful to the body
300 mg/L : Excellent
(300 − 600)mg/L : Good
(600 − 900)mg/L : Fair
(900 − 1200)mg/L : Poor
Above 1200mg/L : Unacceptable
3.3.4 Biological Oxygen Demand : BOD is another prominent water quality calculationterm.Awatersupplywith a BODlevel
of 3-5 ppm can be considered as moderate, with a BOD level of 6-9 ppm water is considered contaminated because there is
usually an organism present and germs decompose the waste. When BOD levels are100 ppm or more, the water supply is
considered to be highly contaminated by organic waste. The need for biochemical oxygen or the amount of oxygen needed to
decompose organic matter per one liter of polluted water. The more polluted the water, the moreBODwill bebecausemostof
it will be organic matter and, as a result, more oxygen will need to decompose. It is therefore a reliable gauge for the pollution
of large amounts of water. One of the main reasons for treating contaminated waterbeforeit isdischargedfroma watersource
is to lower its BOD — that is, reduce its oxygen demand and thus reduce its need for the water from which it is released.
3.3.5 Dissolved Oxygen: Melted oxygen (DO) is a measure of how much oxygen is dissolved in water - the amount of oxygen
found in aquatic organisms. The amount of oxygen dissolved in a stream or pond can tell us a lot about water quality. Melted
oxygen saturation is reported in units of mg / l (mg / l is also called in fractions per million (ppm) because it is 1000 grams of
pure water, and a milligram is a fraction of a million of that). Percentage filled space is reported by percentage units. The
oxygen dissolves in the water until it is full, the normal amount of heat given. Percentage gain tells us which part of the catch
capacity is actually taken. The high level of DO in a public water supply is good because it makes drinking water taste better.
However, high levels of DO accelerate corrosion in water pipes.
3.3.6 Phosphate: Phosphates are the chemicals which contain the element phosphorus,andtheyaffectthequality ofwaterby
causing algae growth in a huge amount. Phosphorus naturally occurs in rocks and other minerals. During the natural climate
process, rocks slowly release phosphorus as phosphate ions dissolve in water and mineralized phosphate compounds break
down, Phosphates
PO4-3 are formed from this component. Phosphates basically exist in three forms: orthophosphate, metaphosphate (or
polyphosphate) and organically bound phosphate; each compound contains phosphorus in a differentchemical system.These
types of phosphate occur in fossils of living and decaying plants, such as free or slightly ion-binding ions in aqueous systems,
chemically synthesized in structures and soils, or as mineralized compounds in soils, rocks and objects.
3.3.7 Nitrate: Nitrate can also cause adverse health affects when present in high quantity. The EPA has established an official
drinking water standard of 10 milligrams of nitrate per liter of water (10 mg / L). The question arises how does nitrate affect
aquatic life? Nitrates have the same effect on aquatic plant growth as the phosphates as well as the same adverse effect on
water quality. Plants and algae grow fast which provides fodder for fish which could lead to an increase in the number of fish.
The demerit of it is that the oxygen levels in the water will drop and the fishes will die.
Because nitrate can be short-lived in a modified form of nitrites, and because nitrites can cause serious illness in both wildlife
and humans, the acceptable nitrate levels in drinking waterhave beendetermined as10mg/l.Pollutedwaterusuallycontains
nitrate less than 1.0 mg / l.
3.3.8 Turbidity : Turbitity measures the relative clarity of the water .It is defined as the number of suspended particles in
water . Debris in the water is caused by organic matter such as clay, mud, dustandorganicmatteraswell asplanktonandother
microorganisms that interfere with the passage of light coming into the water. Turbidity can be referredtoastotal suspended
solids (TSS) and it also involves plankton and other organisms. Natural water variations tend to rise during heavy flow due to
increased ground runoff, river flow, and erosion. High turbidity can disturb the cleanliness and give an environment for
bacterial growth as well as can increase the presence of microbes in water.

3.3.9 Fecal Coliform :The existence of fecal coliform bacteria in fluvial areas designate that water is infected with human or
animal feces. Whenever this happens, the source water is likely to be infected with germs or pathogens thatmaybe presentin
the soil.
3.4 Water Quality Index A Water Quality Index (WQI) is a methodology where water quality is calculated andanalyzed.
Just like Air Quality Index and UV Index ,it calculates index lyinginparticularrangeand describeswhetherthe qualityof water
is safe for drinking or not and ultimately finalizes the result by scaling the water in different scores .
completed. Field values can be converted to index values; respondents were asked a series of questions to include a water
quality level graph (0 to 100) corresponding to field values suchaspH(2-12).Thecurvesarethenweighedandare designedto
calculate the optimum result . The calculator completes individual and group values and allows the respondents to generate
custom reports of the water . WQI calculator will eventually create scaling of the river water tested .
In this study, the calculation of 9 analyzed physicochemical parameters i.e.. PH, Temperature,SolidSolidCore,SolubleOxygen,
Nitrate, Phosphate, Biological Oxygen Demand, Turbidity and Fecal Coliform were selected to assess the quality of the Ganga
River in three different periods ( Summer, Monsoon, Winter) on selected sites.
The water quality indicator is a 100-point scale that summarizes the results from a total of nine different scales when

The Q-value for each test should be multiplied by the rating scale shown on the Worksheet for each test, and the answer
should be recorded in the “Total” column. The measurement factor indicates the importance of each test for overall water
quality. For example, the weighting factor of fecal coliform is 0.16, so it is considered more important in assessing total water
quality than nitrate, with a measuring factor of only 0.10. Finally, add the numbers shown in the Content column to determine
the total Water Quality Index (WQI) of the tested water source. Compare your Index result with the ratio shown in Table I to
measure the water level of the tested water supply
The algorithm used in this study is "Decision Tree Classifier" so the water quality aftercalculating9parametersis determined
by categories and eventually classified into varied classes on quality scaling. The Water Quality Index Calculatorusesa scaling
from 0 to 100 to measure water quality, where 100 is the highest score.
4. RESULTS
Information about river water quality is important for the conservation and survival ofmarineanimals.People,somehow,rely
on rivers to meet their daily needs and human intervention in rivers makes water levels worse. To maintain this perspective,
the current study was designed to assess the surface water level of the Ganga RiverusingMLApplicationstoIoT.Inthepresent
study, nine physicochemical parameters were analyzed. Thealgorithmusedinthisstudyis"DecisionTreeSeparator''sowater
quality after calculating 9 parameters is determined in stages. The Water QualityIndexCalculator usesa scaling from0to100
to measure water quality, where 100 is the highest score. Once all the WQI points are known, they can be collectively used to
determine how healthy the water is on a particular day.
WQI Scale : CATEGORY QUALITY
91 − 100 : Excellent Water Quality
71 − 90 : Good Water Quality
51 − 70 : Medium or average water Quality
26 − 50 : Fair Water Quality
0 − 25 : Poor Water Quality

4.1 Accuracy Score
Accuracy is a metric for the evaluation of classification models. It is the fraction of predictions our model calculated right.
Accuracy = Number of Accurate Predictions / Total Total Predictions
For binary calculations, accuracy can also be calculated according to the positives and negatives as shown below:
Accuracy = (TP + TN) / (TP + TN + FP + FN)
where TP = True Positive, TN = True Negative, FP = False Positive, and FN = False Negative
5. CONCLUSION
The present study investigates the quality of the surface water of the Ganga River in the active river mine area of Varanasi,
Kolkata, Jahangirpur, and Tribeni. The study concluded on the basis of the ML Algorithm accuracypointsindicatingthequality
of water that was slightly polluted during heavy rainfall. at all sample sites. During the summer, WQI scores indicate an
acceptable level of water quality and during the winter, it reflects the excellent water quality of the Ganga River. The water
level of the Ganga River is recorded and polluted during storms due to high runoff from the river compared to summer and
winter seasons. During the WQI quality inspection, at least a good condition was recorded during heavy rains. The study
revealed that the WQI of the Ganga River was not found to be suitable for drinking purposes during the rainy season, itmaybe
suitable for irrigation purposes. The outcome of this current study requires careful monitoring ofthe ecological featuresofthe
aquatic environment especially in the active Ganga rivers due to the potential natural hazard.
5.1 Know Your Water Quality We have made a website named KNOW YOUR WATER QUALITY whichwill predict thequality
status of water ,all we have to do is to feed the details of 9 parameters required for WQI calculation.
LINK:https://gangawaterproject.herokuapp.com/

6 . FUTURE SCOPE
 Accuracy Score can be further improved and more parameters can be included in the WQI Calculation for the better
results.
 A study and a website can be made for many rivers of India together .
 The future scope of this study is that it can be further deployed on Django rather than Flask at this time so that User
Experience can be more interactive and an app can also be made for the same so that one can use it on their
respective smartphones.
 As AQI is a prominent term due to high pollution in metropolitan cities ,Similarly WQI should also be in more
consideration for the riverine systems.
REFERENCES
[1] A.K. Bisht, R. Singh, R. Bhutiani and A. Bhatt, “Artificial Neural Network Based Water Quality Forecasting Model for Ganga
River, ” International Journal of EngineeringandAdvancedTechnology(IJEAT)ISSN:2249-8958(Online),Volume-8Issue-
6, August 2019, pp 2778-2785
[2] Y. Soni and V. Sejwa, “An Estimating Model for Water quality of river Ganga using Artificial Neural Network,”International
Journal of Innovative Technology and Exploring Engineering (IJITEE) ISSN: 2278-3075, Volume-8 Issue-9, July 2019, pp
1448-1453
[3] A.K. Bisht, R. Singh, R. Bhutiani, A. Bhatt and K. Kumar, Water quality modeling of the River Ganga using artificial neural
networks with reference to the varioustrainingfunctions,”EnvironmentConservationJournal 18(1&2)41-48,April 2017,
pp-41-48.
[4] A. K. Haritash, S. Gaur and S. Garg,” Assessment of water quality and suitability of River Ganga in Rishikesh, India, ” Appl
Water Sci (2016) 6:383–392 DOI 10.1007/s13201-014-0235-1, pp 382-392
[5] S.Vijay and Dr.K.Kamaraj, “Ground Water QualityPredictionusingMachineLearningAlgorithmsinR”,IJRAR-International
Journal of Research and Analytical Reviews, e ISSN 2348 –1269, VOLUME 6 I ISSUE 1 I JAN. – MARCH 2019, pp 743-
BIOGRAPHIES
IIT BHU and is currently working on Python and Machine Learning.
Ankur Kashyap : He is a student of B.Tech FinalYear.DepartmentofComputerScienceEngineering,Rameshwaram
Institute of Technology and Management and currently working on Python and Machine Learning projects.
Sakshi :She is a student of B.Tech Final Year,Department of Computer Science and Engineering ,Rameshwaram
Institute of Technology and Management and currently working on Python and Machine Learning.
Kaushiki Agrahari : He is a student of B.Tech Final Year.Department of Computer Science Engineering
,Rameshwaram Institute of Technology and Management . She has completed an internship on Machine Learning from

Shashank Kumar: He is a student of B.Tech Final Year.Department of Computer Science Engineering
,Rameshwaram Institute of Technology and Management and currently working on Python and Machine
Learning projects.
Shivansh Sharma: He is a student of B.Tech Final Year.Department of Computer Science Engineering ,
Rameshwaram Institute of Technology and Management and currently working on Python and Machine
Learning projects.
Shyam Dwivedi –He is currently working as an Assistant Professor and Head of Department in Rameshwaram
Institute of Technology & Management, Lucknow, India. He is M. Tech – 2012 BIT Mesra, Ranchi, he has a teaching
experience of 10 years and 1-year in TCS Industrial experience.
.

Water Quality Index Calculation of River Ganga using Decision Tree Algorithm

Recommended

More Related Content

Similar to Water Quality Index Calculation of River Ganga using Decision Tree Algorithm (20)

More from IRJET Journal (20)

Recently uploaded (20)

Water Quality Index Calculation of River Ganga using Decision Tree Algorithm