Description
Assignment #4
CSCI 5408 (Data Management, Warehousing, Analytics)
Problem #1
Business Intelligence Reporting using Cognos
1. Download the weather dataset available on https://www.kaggle.com/PROPPG-PPG/hourlyweather-surface-brazil-southeast-region?select=sudeste.csv
2. Explore the dataset and identify data field(s) that could be measured by certain factors or
dimensions. (Follow recorded lecture #18, and synchronous session #18)
Example: In a Sales dataset, you may find a measurable field “total sales”, which could be
analyzed by other factors such as, “products”, “time”, “location” etc. These factors are known as
dimensions. Depending on the data, you may also find possibilities of slice and dice, i.e. analysis
could be possible in more granular level; From total sales by city to total sales by store
3. Write ½ page explanation on how did you select the measurable filed, i.e. fact and what are the
possible dimensions. Include this part in your PDF file.
Summer 2021 saurabh.dey@dal.ca
4. Clean the dataset, if required perform formatting. You can perform the cleaning and formatting
using spreadsheet operation or programming script. If you use program add that in GitLab, if
you use other methods, write the steps in the PDF file.
5. Create Cognos account and import your dataset. Identify the dimensions, and create/import the
dimension tables.
6. Based on your understanding of the domain (please read the information/metadata available
on the dataset source, i.e. Kaggle), create star schema or snowflake schema. Provide justification
of your model creation in the PDF file.
7. In addition to justification, attach screenshot of the model (star schema or snowflake schema)
in the PDF file.
8. Display visual analysis of the data in a suitable format, e.g. bar graph showing temperature
change in terms of a suitable dimension. Add the screenshot of the analysis on the pdf or add a
screen recording of the analysis on your .zip folder.
Problem #2
Sentiment Analysis – Java Program only
1. To perform this task, you need to consider the processed news (“content or descriptions” only,
ignore other fields) that you obtained and stored in MongoDB in your previous assignment. If you
could not perform/complete the task, then obtain the processed MongoDb News collection by
contacting your TA Kethan (Cc me in that email)
2. Write a script to create bag-of-words for each news article. (code from online or other sources
are not accepted)
e.g. news1 = “Canada is cold cold. I feel good not bad”
bow1 = {“Canada”:1, “is”:1, “cold”:2, “I”:1, “feel”:1, “good”:1, “not”:1, “bad”:1}
You do not need any libraries. Just implement a simple counter using loop.
Compare each bag-of-words with a list of positive and negative words. You can download list of
positive and negative words from online source(s). You do not need any libraries. Just perform
word by word comparison with a list of positive and negative words that you can get from any
online platform. E.g. negative words can be found here https://gist.github.com/mkulakowski2/4289441
3. Tag each news as “positive”, “negative”, or “neutral” based on overall score. You can add an
additional column to present your finding.
E.g. frequencies of the matches “cold”=2, “not” =1, “bad”=1 (negative -4), “good”=1 (positive +1). Overall score = -3
News Article News
Description/content
match polarity
1 Canada is cold cold. I
feel good not bad
cold, good, not, bad negative
Summer 2021 saurabh.dey@dal.ca
Problem #3
Semantic Analysis
1. For this task, consider the processed news collection that you created in Assignment 3.
2. Use the following steps to compute TF-IDF (term frequency-inverse document frequency)
a. Suppose, you have 50 news articles (description or content only) that are stored in
50 JSON arrays. You need to consider these data points as the total number of
documents (N). In this case N=50
Now, use the search query “Canada”, “Moncton”, “Toronto”, and search in how many
documents these words have appeared.
Total Documents 50
Search Query Document
containing
term(df)
Total Documents(N)/ number
of documents term appeared
(df)
Log10(N/df)
Canada 30 50/30 0.221848749
Moncton 5 50/5 1
Toronto 10 50/10 0.698970004
b. Once you build the above table, you need to find which document has the highest
occurrence of the word “Canada”. You can find this by performing frequency count of
the word per document.
Term Canada
Canada appeared in 30 documents Total Words (m) Frequency (f )
Article #1 6 2
Article #2 10 1
: : :
Article #30 8 1
c. You should print the news article (programmatically), which has the highest relative
frequency. You can find this by computing (f/m).
Assignment 4 Submission Format:
1) Compress all your reports/files into a single .zip file and give it a meaningful name.
You are free to choose any meaningful file name, preferably – BannerId_Lastname_firstname_5408_A4
but avoid generic names like assignment-4.
2) Submit your reports only in PDF format.
Please avoid submitting .doc/.docx and submit only the PDF version. You can merge all the reports into
a single PDF or keep them separate. You should also include output (if any) and test cases (if any) in the
PDF file in the proper format (e.g. tables, charts etc. as required).
3) Your executable code/script needs to be submitted on https://git.cs.dal.ca/
