Executive Summary

In every solution understanding the problems is vital before taking any actions, likewise,
understanding problems in organisation and explore data in meaningful ways is the
contribution of data analysis. This report explore how we find a raw dataset which may
contain meaningful patterns by using our analysis skills, after taking necessary steps we
proceed to analysis the data and visualize them using analytical tools. Similarly, report covers
the temperatures record in different State of Brazil from 1990 to 2019, various finding
regarding descriptive, predictive, and prescriptive analysis has made by utilizing the skills and
knowledge we have gather so far through this course.
Table of Contents
Executive Summary .......................................................................................................... 1
Table of Figures ................................................................................................................ 2
Introduction ..................................................................................................................... 3
Purpose ................................................................................................................................... 3
Scope ...................................................................................................................................... 3
Definition of terms ................................................................................................................... 3
Data Analysis Findings ...................................................................................................... 4
Review of literature related to similar dataset .......................................................................... 4
Background ................................................................................................................................................... 4
Discussion ..................................................................................................................................................... 4
Conclusion .................................................................................................................................................... 4
Summary of Groups Findings .................................................................................................... 4
Possibilities in Dataset .................................................................................................................................. 4
Problems in Dataset ...................................................................................................................................... 5
Discussion about finding including models and Visualizations to support the findings ................ 5
Descriptive Analysis ..................................................................................................................................... 5
Predictive Analysis ..................................................................................................................................... 10
Prescription Analysis .................................................................................................................................. 12
Conclusion/recommendations ........................................................................................ 14
References...................................................................................................................... 15

Table of Figures
Figure 1 Map of Brazil with 5 states .......................................................................................... 5
Figure 2 Annual Average with Clustering ................................................................................. 6
Figure 3 Average Annual Min/Max ........................................................................................... 6
Figure 4 Annual Average each state .......................................................................................... 7
Figure 5 Each state annual average clustering ........................................................................... 8
Figure 6 Polygon ........................................................................................................................ 8
Figure 7 Sample of Para ............................................................................................................. 9
Figure 8 Annual forcasting ...................................................................................................... 10
Figure 9 Forcasting for each state ............................................................................................ 10
Figure 10 Revenue data with clustering................................................................................... 12
Figure 11 Revenue data ........................................................................................................... 13
Figure 12 Revenue forecasting ................................................................................................ 13
Introduction
Data analytics is the process of examining datasets to draw conclusion about the information
they contain, Green, Willis, Hughes, and et al, 2007. In the process of data analytics, initially
the raw data will be selected, or businesses organization stored data in raw form. In our case,
we select raw dataset from ‘Kaggle.com’, and the data set is about temperature measured in
different state of Brazil in different years and months. After selecting dataset, we proceeded
to data cleaning process, data may contain duplication or irrelevant observation which needs
to remove, similarly that the structure of our dataset should be fixed and should handle
missing data. After the cleansing process, dataset is ready for analysis, we select Tableau as
our prior data analytic tool. In addition, for the purpose of prescription analysis, we merge
another dataset of ‘Ice-cream Parlour Revenue’. Finally, we are able find different patterns
and show them as a chart and dashboard.

Purpose

The purpose of the assessment is to implement different data analytics skills we have
developed so far in this course. This assessment allows student to search for dataset on their
own which empower the student’s ability to analyse the raw set from where we can find some
interesting patterns. Not only this, but it also allows us to use professional tool like Tableau
and Excel from where students can discover more features of those tools and utilities
theoretical knowledge that have into practical. Most importantly, this assessment facility
student to learn each step of data analytics process including the presentation which helps us
on how we can expose our finding to others for the purpose of knowledge sharing
or making any business decision.

Scope

• Exploring of different professional dataset

• Utilization of skills and knowledge by using professional tools
• Group activity for better practice of working as a team than as an individual
• Enhance the ability of analytical skills
• Visualization and dashboard for problem solving

Definition of terms

a. Seasons in Brazil
• December, January, and February (DJF) are Summer.
• March, April, and May (MAM) are Autumn
• June, July, and August (JJA) are Winter
• September, October, and November (SON) are Spring

b. Temperature is in Celsius. For example: When we write average temperature 26.6 that
means it’s 26.6 degree Celsius.
c. All the date are in ‘AD’. For example: when we write 2019 then it means 2019 AD.
Data Analysis Findings

Review of literature related to similar dataset

Background

Before starting the analysis process, we research on similar dataset that other had user for
their analysis process. We find “District Wise Rainfall in India” dataset, this dataset also
contains similar structure and datatypes like ours. This dataset mostly describes the density
of rainfall in different district of India.

Discussion

The structure of this data is like ours and many of professionals have already use this dataset
for the analysis purpose. This helps us on finding what could be the possibility of our dataset.
There are lots of possibilities which can explain the geographical reasons, climate, land
suitable for agricultural purpose and other much information are included on this data set.

Conclusion

At last, we get much information how to handle the structure and missing value in the data
from our research on this dataset. In addition, we also learn what sort of analysis will be
appropriate for our dataset.

Summary of Groups Findings

Possibilities in Dataset

The dataset is about temperature records in different months of the years from 1990 to 2019
in 5 states in Brazil. From the dataset we have discover geographical description according to
state. Similarly, we divide our finding into temperature of Months, Seasonal temperature (4
season in Brazil) and Annual Temperature. In addition, we can take one state as a sample and
compare those findings with how it is different than the combination of all 5 state.

Similarly, for the purpose of prescriptive analysis we choose another data set which describe
relationship between the revenue generated by Ice-cream Parlour and the temperature. In
the case of Ice-cream Parlour, we found that the relation of revenue and temperature is
directly proportion. We show the average temperature in “Temperature in different state of
Brazil” dataset to temperature of “Ice-cream Parlour Revenue” dataset. Finally, we can
prescribe new investor who is willing to inverts in Ice-cream parlour to choose which state
can give him more profit.

Note: Assuming that temperature is vital components effects sale and ignored other factors
in the prescription analysis.
Problems in Dataset

From analysing the dataset, we had discovered two major problems which will directly affect
the data analysis process.

a. Structural Problem: The dataset we have selected contains different excel file for each
state. For the intension of best finding, we merged the dataset which fix our issues on
structure of dataset. Similarly, we delete the repeated columns (where required).

b. Missing Values: For the missing value, the dataset by default had replaced missing
value by ‘999.9’ but the problem with that is temperature cannot be on that level
which results in inaccurate analysis of the data. So, we replace all the ‘999.9’ values
with average temperature of that month through the available year.

Discussion about finding including models and Visualizations to support the findings

According to the type of analysis, we also divide our analysis on three parts i.e. Descriptive
analysis, Predictive Analysis and Prescription Analysis

Descriptive Analysis

Descriptive Analysis is the type of analysis of data that helps describe, show, or summarize
data points in a constructive way such that patterns might emerge that fulfill every condition
of the data (Zeng, Fu, Arisona, S.M. and Qu, 2013). Firstly, we identified different patterns
that the data describe using different plots and charts. The first description is about the
geography, we have located Brazil and its different State as a geographic Map, also indicate
the hottest and the coldest State in Brazil differentiate by the colour. From the Map we
identified that Parana and Rio de Janeiro State is comparatively coldest area where
Amazonas, Para and Amapa is comparatively hottest state, one of the reasons of these three-
state having hot climate is because they are closer to equator line.

Figure 1 Map of Brazil with 5 states

From the bar graph with clustering, we can clear identify that highest average annual
temperature is recorded in 2015 where lowest is in 1996. Bar graph also shows that last 6-
year average temperature is comparatively maximum than any other year before.

Figure 2 Annual Average with Clustering

According to the Min/Max line graph, from 1990 to 2019, lowest temperature recorded is in
1999 which is 18.2. Similarly, in graph we can see high fluctuation rate in Minimum average
where there is steady line in Maximum average temperature.

Figure 3 Average Annual Min/Max

Average annual according to state and season

There are four seasons in Brazil, we use parameter to divide months into seasons and assign
calculative field for making season parameter work. Through horizontal bars, we can see
Amazonas is always the hottest and Parana is lowest every season. However, average
temperature in Para, Rio de Janeiro and Amapa fluctuate according to season. For example:
Rio de Janeiro’s average temperature in Summer is 27.3 degree which is higher than Amapa,
but in Autumn it falls to average of 21.68 degree. In the other hand Amapa remains on similar
temperature and became 2nd hottest state in Autumn. The records shows that in Rio de
Janeiro most temperature changes, it became much hotter in summer and much colder in
winter. Pie-chart has shown similar result much clear, where Amazonas takes maximum size
with average temperature in autumn is 28.58 where Parana with average temperature of
15.46.

Figure 4 Annual Average each state

Figure 5 Each state annual average clustering

Annual average by month from 1990 to 2019

From the Polygon we can find some of the interesting information. In the month of January,
March, April, and October almost always the temperature is below the trendline where
December is the only month that is almost always higher than the trendline in recorded
average temperature from 1990 to 2019

Figure 6 Polygon
Sample of para

We have taken Para as sample state for the comparison of how each state is different with
annual temperature recorder from year 1960 to 2019. Para is also one of the hottest states
among the 5. The average temperature of Para is around 26 degrees, so we can say that
there is tropical savanna climate. The hottest average temperature recorder is 29.070 on
the month of November in 2017 where lowest recorder average temperature is 25.410 in
1962.

Figure 7 Sample of Para

Predictive Analysis

Predictive analytics is a branch of advanced analytics that makes predictions about future
outcomes using historical data combined with statistical modelling, data mining techniques
and machine learning, Nithya, and Ilango, 2017. To predictive analysis, we have excluded 2019
from the data set (to verify the accuracy of the forecasting) using filter and has forecast for
next 10 year, which is 2028.

Annual average prediction

According to the line graph, the average annual temperature recorded in 2018 is 25.959 which
is expected to be 26.125 and 26.368 in 2023 and 2028 respectively. This shows that the
temperature of Brazil is in increasing trend up to 2028.

Figure 8 Annual forcasting

Annual average prediction by season

Some of the finding on forecasting on annual average by different season has been displayed
as table.

Season\Year 2018 2023 2028

Summer 26.511 26.989 27.206
Autumn 24.867 25.191 25.394
Winter 26.054 25.866, 26.257
Spring 26.404 26.524 26.615

Figure 9 Forcasting for each state

Annual average prediction of each State

Like always, Amazonas will remain the hottest and Parana will remine the coldest, this
position will continue up to 2028 according to the forecasting. However, Amapa temperature
will slightly increase than Para after 2020.
Prescription Analysis

Prescriptive analytics is a type of data analytics by use of technology to help businesses make
better decisions through the analysis of raw data (Appelbaum, Kogan, Vasarhelyi, and Yan,
2017). For the prescription analysis, we select another dataset of Ice-cream parlour average
revenue corresponding to the temperature. In the data set there was 2 columns, one is the
temperature and the another is average revenue per day. For Prescription analysis, we have
assumed that we have third person who wants to invest in new Ice-Cream shop in Brazil, but
he is in dilemma which state is suitable. So, from the analysis we are trying to give him
suggestion what State will be the best for better outcome. Not only this, but we have also
utilized forecasted temperature to make the decision on how the revenue will generated in
2019 and how it’s gone change in 2022 according to which state s/he wants to invest.

Figure 10 Revenue data with clustering

Figure 11 Revenue data

Figure 12 Revenue forecasting

Assumptions: If the opening day of the shop is 320 and other dependency of revenue has
negated.

From the above figure we can find that, Para and Ampa state had highest revenue in 2019
followed by Amazonas. However, in 2022, Amazonas will be the first to generate highest
revenue where Para and Amapa will remain the same as in 2019. This shows that Amazonas
is better state to inverts.
Conclusion/recommendations
Analysis of the data helps finding various secrete which can be use as information for any
decision taking process. In the assessment, we successfully completed each process of
analysing the data. Initially taking raw data contains very less and inaccurate information.
Cleansing process and restructuring data set is crucial part of or data analysis process. From
the temperature dataset of Brazil, we can determine the hottest and coldest region, different
flection over the years. Also, we can get information on different seasons in Brazil and
temperature change over last 30 years. In addition, we can predict the future temperature of
different geography in different season. We are also able to forecast the temperature up to
2028 on the basic of average annual, different seasons, different state, and months. From the
comparison we took 2 different dataset and established relationship for the purpose of
prescription analysis. Prescription analyses help us in taking business decision and can play
vital role in decision making process.

Lesson we learned: Varieties of columns in dataset will gives different and accurate analysis.
In the future we would like to suggest using more complex dataset to enhance the analytical
skills and to find more insights of the dataset. Verities of charts, tables, bars and etc are useful
to understand the multiple dimensions of the dataset. Finally, I will suggest using more
insightful dataset in future. Suggestion from lecture and peers helps us in understanding the
different prospective of analysis.
References
Appelbaum, D., Kogan, A., Vasarhelyi, M. and Yan, Z., 2017. Impact of business analytics and enterprise
systems on managerial accounting. International Journal of Accounting Information Systems, 25, pp.29-44.

Green, J., Willis, K., Hughes, E., Small, R., Welch, N., Gibbs, L. and Daly, J., 2007. Generating best evidence from
qualitative research: the role of data analysis. Australian and New Zealand journal of public health, 31(6),
pp.545-550.

Nithya, B. and Ilango, V., 2017, June. Predictive analytics in health care using machine learning tools and
techniques. In 2017 International Conference on Intelligent Computing and Control Systems (ICICCS) (pp. 492-
499). IEEE.

Zeng, W., Fu, C.W., Arisona, S.M. and Qu, H., 2013, June. Visualizing interchange patterns in massive movement
data. In Computer Graphics Forum (Vol. 32, No. 3pt3, pp. 271-280). Oxford, UK: Blackwell Publishing Ltd.