- [2.5 points]. Run a thorough univariate and bivariate graphical and statistical examination of your data. Do you notice any irregularities in your data? How does your data look like? Normal, skewed?
Tips. Make sure to number and label each table and graph. (e.g., Table 1. Summary statistics for HBA missing data set).
Provide a title and a detailed interpretation of each chart.
Remember the rule of thumb: for any table and/or graph, you can have at most 3 genuine findings. If you have more, you have probably made them up (����). - [2.0 points]. Missing values analysis. Do you have any missing values in your data? If
yes, determine the extent of missing values per variable and case. Are there any
variables/cases that you need to delete? Use ≥ 30% of missing values as the threshold
for deletion.
a. After deleting variables and/or cases with ≥ 30% missing data, construct a
summary statistic for your data. Do you still observe any missing values? If yes,
decide on how to impute these missing values. Limit your imputation
technique to mean or median substitution. Justify your choice. - [2.0 points]. Detection and treatment of outliers.
Math and Statistics for Analytics Nov 2020
3
Are there any univariate outliers in your dataset? Use both the Tukey´s fences and z
score approach (with z threshold set at 2.5 since you have a small sample size) to
identify them. Do you notice any discrepancy between the two methods? Explain.
How many values were detected as outliers? Will you keep these outliers or delete
them? Justify your decision. Discuss impact of your decision on remaining data
analysis. - [2.5 points]. After treating missing values and outliers, construct a summary statistic of
your data. Compare and contrast your results with question 1. Develop two
hypothetical questions that you can answer using graphical and/or empirical
techniques. Provide correct answers. - [1.0 point]. In the case that you did not treat missing values and/or outliers, what
would be the impact on subsequent data analysis.
Dataset: HBAT Industries Dataset – HBAT missing. (You can access the dataset in Additional
Documentation / Assignment 1)
Context: HBAT is a manufacturer of paper products. Hypothetical dataset based on surveys of
HBAT customers completed on a secure web site managed by an established marketing
research company.
Sample size: 70 observations on 14 separate variables based on a market segmentation study
of HBAT customers: newsprints industry and the magazine industry.
Categories of data:
• Numerical variables: V1 to V9.
• Categorical variable: V10 to V14.
Additional information related to the variables is available in the excel file (HBAT missing) in
the Metadata spreadsheet.
Pre-requisite: before working on this assignment, you need to watch the series of videos: Data
examination – Excel (Campus Online / Additional Documentation / Module 2 / Recorded
Videos)
For detecting outliers, you are already familiar with the boxplot method as well as the Tukey´s
fences (video), you can also use the z score approach, to calculate the z value for each
observation, you can use Excel built-in function (STANDARDIZE).
For missing value analysis, you can use the COUNT function to count the numbers of cells
containing data in a range that contains numbers and use it to determine the extent of missing
values per case and per variable.
Additional resources:
Introduction to data analysis in Excel: https://www.youtube.com/watch?v=Rs4082ewxgA
Introduction to Pivot tables: https://www.youtube.com/watch?v=9NUjHBNWe9M
Math and Statistics for Analytics Nov 2020
Welcome to Legit Writing
