Indicating Ethanol Quality Using Machine Learning

R.Deeptha1, Anish G2, Johith Erthineni3, Gouthamsai Gundam4

1,2,3,4
Department of Information technology, SRM Institute of Science & Technology,
Ramapuram campus, Chennai

E-mail: [email protected], [email protected],

[email protected] ,[email protected]

Abstract- Both customers and the ethanol

business industry about the ethanol's quality[1]. It
takes time to measure ethanol purity the II. RELATED SURVEY
conventional (professional) technique[2]. Machine
learning models are crucial tools today for
replacing human work[3]. There are a number of Moderate ethanol intake reduces pressure and will
features that can be used to estimate the ethanol increase emotions of happiness and well-being, and
quality in this scenario, but they are not all can lessen the chance of coronary heart disease.
necessary for an accurate prediction[4]. Heavy intake of alcohol, however, might also
Therefore, the focus of our thesis is on the additionally reason dependency and will increase
characteristics of ethanol that are critical to all varieties of damage and trauma. Environmental
achieving a promising outcome[5]. We employed and genetic elements are worried in susceptibility to
three algorithms—Random forest, Decision tree, alcoholism. Ethanol can result in malnutrition, and
and assessment—for the classification model and may exert an instantaneous toxicological impact
evaluation of the pertinent features[6]. Dry because of its interference with hepatic metabolism
Ethanol and Methyl Ethanol were the two ethanol and immunological functions. A causal impact has
quality datasets we used in this investigation[7]. been determined among alcohol and numerous
We used the Pearson coefficient correlation to cancers. Cessation of alcohol intake and balanced
examine the importance of the property[8]. nutrients are encouraged number one nonspecific
healing measures for alcoholics. Drug cures for
Keywords: Ethanol Quality, Indicating Ethanol alcoholics affected by damage has ended in
Quality, Machine Learning (ML) Project, combined results. In end-degree liver disease, liver
transplantation can be considered. As a result, they
can gain a better of the quality of the ethanol, which
I. INTRODUCTION obviously saves a tone of money and time.
Furthermore, this assisted in gathering a tone of
data on many parameters, including the quantity of
The most widely used substance on the planet is various chemicals and temperature employed
ethanol, and society values it highly. In the current during production, as well as the caliber of the
cutthroat market environment, ethanol quality is ethanol produced. Numerous databases include
crucial for producers as well as consumers. Testing at these data (UCL Machine Learning Repository, and
the end of manufacturing used to be the traditional Kaggle). Various efforts have been made to
method for determining the quality of ethanol; to determine the quality of the ethanol using the data
reach that level, one must already invest a significant currently available, with ML techniques' success in
amount of time and money. If the quality is poor, the past decade contributing to this. One can adjust
various procedures must be implemented from the variables that directly affect the quality of the
scratch, which is quite expensive. It can be difficult ethanol during this process. This gives the
to determine a quality based just on a person's manufacturer a better notion of how to adjust
laboratory tests because each person has a different various development process factors to adjust the
perspective about the quality. As technology ethanol quality. Therefore, it is crucial to analyze
advanced, manufacturers began to rely more and the fundamental factors that impact ethanol quality.
more on various devices for testing during the
development process.
 III. SYSTEM MODULES

A. Random Forest:

A supervised learning technique called random forest

uses a random decision tree in conjunction with
classification and regression. They work by building
numerous decision tree layers during training, and
they output the classes or mean prediction of each
tree. The total probability is now obtained by adding
up the probabilities of all the decision trees. It is
mostly applied to high-dimensional data sets when
the individual variables are excessively noisy and
non-stationary. That's because, like a forest, it is
made up of several decision trees. Fig 1: Random Forest Representation

Additionally, it employs randomness to improve its B. Train Test Splitter:

precision and prevent over fitting, which can be a
major problem for such a complex algorithm. Based When machine learning algorithms are used to make
on a random selection of data samples, these predictions on data that was not used to train the
algorithms create decision trees and obtain model, their performance is estimated using the train-
predictions from each tree. They then vote to test split technique. It is a quick and simple process to
determine which viable option is the best. It has carry out, and the outcomes let you compare the
several uses in daily life, including image classifiers, effectiveness of machine learning algorithms for your
recommender systems, and feature pickers. Its real- particular predictive modelling issue. Despite being
world applications include sickness prediction, loan straightforward to use and understand, there are some
application classification, and fraud detection. It circumstances in which the method shouldn't be
serves as the foundation for the Boruta algorithm, applied, such as when the dataset is tiny and further
which selects crucial dataset properties. configuration is needed, as when it is used for
classification and the dataset is unbalanced. Based on
The random forest will select "k" features randomly KNN algorithm for machine learning helps for
from your dataset's "m" features. By selecting the decision support, quality improvement, health care
node with the maximum information gain among the cost reduction, diagnosis, treatment process designing
k characteristics, the algorithm will now calculate the and remote monitoring. It automatically helps the
root node. The algorithm then divides the node into healthcare professionals to identify the real health
child nodes and then repeats. The algorithm then issues of the patients and give effective solution to
divides the node into child nodes, repeating this the problem for the patients which helps the doctor to
operation "n" times. You currently have n trees in a predict the disease and its causes easily and
forest. In order to integrate the outcomes of all the accurately.
decision trees in your forest, you will execute
bootstrapping at the end. C. Decision Trees:

Given that it expands on the capabilities of decision A tree has many similarities in everyday life, and it
trees, it is unquestionably one of the most complex turns out that it has affected both classification and
algorithms. Through an indication of the attribute regression in a broad range of machine learning. A
selection, the algorithm creates the individual decision tree can be used in decision analysis to
decision trees. Every tree is supported by a separate formally and visually reflect decisions and decision-
random sample. Every tree casts a vote in a making. It employs a decision-tree-like approach, as
classification problem, and the classification with the the name suggests. Although a tree is a frequently
most votes wins. In contrast, you would compute the used data mining tool for determining a plan of attack
average of all the tree outputs in a regression to accomplish a certain objective, it turns out that it
problem, and that would be your final outcome has influenced a vast field of machine learning,
spanning both classification and regression. A
decision tree can be used in decision analysis to
formally and visually reflect decisions and decision-
making. As implied by the name, it employs a tree-
like model. Although a common tool in data mining IV. RESULT AND DISCUSSION
for determining a plan of attack to accomplish a
specific objective, it is also widely employed in The detailed analyzed data in the form of the quality
machine learning, which will be the primary topic of of ethanol and by this we are able to indicate the
this article. quality with very high accuracy as shown is the fig 4.

Fig 2: Decision Tree Structure

D. ML Class:

In this model the main objective is to identify an

accurate result of finding the ethanol. So the data is
Fig 4: Values Used
now prepared from the dataset and those data will be
collected by the certain algorithms so the analysis
would be identified also the select algorithm is run
From the dataset entered the detailed analysis of the
and also tested for this model to identify with the
volatile quality that is required to identify the
most accurate result.
accurate quality is shown on the fig 5.

Fig 3: ML Class
 Fig 5: Volatile Acidity VS Quality system that exhibits excellent performance with least
error.

Likewise the detailed analysis for the citric acidity

level that is required to be present in ethanol and
the quality analysis is made sure that is accurate and V CONCLUSION
the analysis is shown is the fig 6.
By implementation we will be able to accurately
indicate ethanol quality using this model and other
machine learning methods. Future effort to indicate
more accuracy than 0.92 with different approaches
and algorithms, and using machine learning model
will be more beneficial with emerging technology.
For this model for indicating the ethanol quality we
would be able to find the approximate values of the
accuracy. When this model is put to use the quality of
the product is set and is very relevant to the updated
dataset helping us find the quality of ethanol. The
model with not only the use of dataset but also will
provide accurate results by using powerful machine
learning algorithms which is random forest and
decision tree making method. Algorithms aid to more
accurate results and possibilities. The accuracy of
random forest classifier is 0.92 and decision tree with
0.82.

Fig 6: Citric Acidity VS Quality

REFERENCES
[1] Quality of Ethanol Produced by Utilization
TABLE 1: Representation of the dataset in this model of Sorghum bicolor by Abdel Moneim Elhadi
Sulieman MA, pp. 853–867.

[2] Ethanol Indicator Based on Instantaneous

Localization by Yu, Yanhao, Soeren Brandt,
Natalie J. Nicolas, and Janna Aizenberg. Support
Syst. 47, 547–553.

[3] Ethanol Degradable Indicator of Hand

Hygiene Quality by Susie Samreth, Rachel
Mosher, Emily Andrews. Open J. Stat. 11, 278–
289

[4] Prediction of the concentrations of ethanol

and acetic acid by Takuo Yano, Tadanori Aimi,
Yasuhisa Nakano, Masahiro Tamai pp. 1–8. Er,
Y., Atasoy, A., 2016.

[5] Taking ethanol quality beyond fuel grade: A

review by Shinnosuke Onuki, Jacek A. Koziel,
The performance of the proposed method is
William S. Jenks, Lingshuang Cai, David
compared with the already reported method by errors.
It is shown in figure 6. The proposed method is Grewell, J. H. van Leeuwen
preferred choice for the data analytics in health care
[6] Evaluation of Ethanol Measuring Technique
by Steven Hennessey, Kevin Payne Estabrooks,
A., Japkowicz, N., 2001.

[7] Deeptha, R. and Rajeswari Mukesh. (2018).

Extending OpenID Connect Towards Mission
Critical Applications. Cybernetics and
Information Technologies. 18(3): 93-110. (ISSN
(Print): 1311-9702, ISSN (Online): 1314-4081).
DOI: 10.2478/cait-2018-0041

[8] Inexpensive Procedure for Measurement of

Ethanol: Roni Miah, Ayesha Siddiqa, Jamsheda
Ferdous Tuli, Noyon Kumar Barman, Shuvra
Kanti Dey, Nihad Adnan, Mamoru Yamada2,
Ali Azam Talukder

[9] Determination of ethanol concentration in

alcoholic beverages by direct analysis in real
time mass spectrometry (DART-MS) by Edward
Sisco, Elizabeth L. Robinson

[10] A Rapid Method for Determination of

Ethanol in Alcoholic Beverages Using Capillary
Gas Chromatography MEI-LING WANG,
YOUK-MENG CHOONG, NAN-WEI SU AND
MIN-HSIUNG LEE

[11] Policy for Testing of Alcohol (Ethanol) and

Isopropyl Alcohol for Methanol by U.S.
Department of Health and Human Services

[12] Detection of Ethanol Concentration using a

Generic Optical Sensor Platform by Abraham
Vázquez-Guardado, J. Alfredo Ramirez-Flores3
Gisela Lopez-Galmiche, J. Jesús Escobedo-
Alatorre, J. J. Sánchez-Mondragón