Criteria
| DSC 4.0 |
Tableau: Level 02 |
| No. of credits |
15 ( subject to DSC Guidelines |
| Mode of training |
Classroom + Doubt clearing through course forum (on demand) |
| Steps to be followed |
- Complete all the pre-requisites mentioned
- Complete all the modules including the assignments
- Submit assignments, if instructed to
- Successfully complete the assessment
|
| Final Assessment Format |
Descriptive Test |
| Assessment duration |
2 hours (Closed book) |
| Pass percentage |
70% |
| Access/installations for taking test |
- Tableau (if personal laptop available)
- Tableau access (Provided by MSU)
|
| No of question paper sets available |
3 sets available on muLearn |
| Does the assessment require answer script submission |
Program PoC to coordinate for descriptive test and other details for answer script submission |
| Mode of score collation |
Program PoC to coordinate for descriptive test |
Final Assessment – Rubrics & Weightage
|
Problem Clarity
(30 Marks)
|
- A muVisual needs to be created and submitted (mandatory). It need not necessarily be on pen/paper, but in form of a flow chart in word/draw.io/freemind
- Check logical flow. For example: Is a master table being created before performing subsets or are there multiple join operations
- Assess the solution with respect to the provided optimized solution as the baseline. The score should be proportional to the number of steps taken by the test taker to perform the operation
- Is the code modularized? Each question and its corollary should have a separate solution
- Assess the complexity of the code by checking the number of loops, control structures, etc. utilized through the O(N) notation. A result obtained through the least complexity is preferred
|
|
Problem Completion
(40 Marks)
|
- Has the data been cleaned and treated? Cleaning involves the removal of NULLs, imputation, etc. Treatment involves polynomial / logarithmic transformations, binning, one-hot encoding as necessary
- Has a univariate analysis been performed, and necessary transformations been applied? Each variable should be tested for summary statistics and should be subjected to normality tests. Non-normal data should be treated as required
- Have appropriate bivariate analyses been performed? Variables can be continuous-continuous, continuous-categorical, or categorical-categorical
- Has a correlation analysis or an appropriate test of association, for example, chi-square test been performed? What is the rationale behind this test? Have the test results been interpreted accurately?
- Is the visualization used appropriately? For example, has a line graph been incorrectly used for categorical variables instead of a bar graph?
- Have all parts of the visualization been completed showcased – chart title, axes labels, data labels, x and y axes, trendlines?
- Code checklist: Is there a QC after each code step, for example, after joins, aggregates, etc.
- Business checklist: Business QC is based on summary statistics of the data. For example, are the forecasted figures in the same ballpark as the moving averages?
|
|
Code Quality
(20 Marks
|
- Functions/libraries should be preferable only when they reduce code complexity and improve performance. This can be ascertained by comparing it with the optimal solution. Note that test takers should not be penalized if they have used a raw execution step which minimizes the complexity of the problem as opposed to using built-in functions
- Was proper naming convention followed while naming data frames and variables? Every major step needs to have a documentation style comment
- Does the number of comments match the number of documentation and single-line comments used in the optimized code? Are there codes for both major and minor steps?
“Three parts to look out for –
a. The code should have an author, edited, purpose etc. section at the top
b. Inputs and output for each module should be provided in the code as comments
c. An additional set of comments should contain details about execution – date run, and time of execution”
|
