Reference no: EM131757175

Data Cleaning

Problem 1 :
The dataset is missing a lot of data, suggest an explanation for

(a) missing payment card numbers

(b) prices 1 and

(c) end addresses

Finally, (d) calculate what percentage of prices are missing and suggest a way to deal with both missing prices and other missing data before using the dataset for analysis

Whatever you decide to do in (d), apply it to the given dataset - including sorting out not missing but inconsistent and missing data data - and save it as stage1.csv in your final submission

Descriptive Statistics
For this problem include in your latex file itself which R-code you used for answering each part of the question.

Problem 2:

a) What is the average journey cost

b) Which journey (between which addresses) is the most popular and what is the median time for this journey?

c) What is the average and median duration of the journeys? Hint: You may find it helpful to create and addition column in excel to hold the calculated journey times, and read the resultant file into R, before doing the analysis.

Modelling

For this problem describe in your latex file itself which R-code you used for answering each part of the question.

Problem 3:

The prices are not missing due to the corresponding transaction being paid in cash 1

a) Create a plot of the relationship between journey time and cost

b) Is there a linear relationship between these variables? Show your reasoning for this answer, mentioning the type of model you use to answer this question

c) Can you suggest a rough set of categories by which journeys can be clustered? Suggest the model that you can use to find this out, and the values representative of the clusters 2. Furthermore, explain what these clusters can be understood to represent conceptually with respect to the journeys

d) Run the appropriate validation test for your clustering model, and explain how this affects your certainty about your categories

Question Refinement and Hypothesis Testing

For this problem describe in your latex file itself which R-code you used for answering each part of the question.

Problem 4 :

As a data scientist hired by Uber you have been asked to simply figure out ways to reduce costs. However, you only have the attached customer data as input. Uber has said that this customer's behaviour is representative of a important sector of the market in the Al Naseem area.

Your task is to figure out if the question of 'how costs can be reduced', can be answered by the given data. Your initial consultation with someone in finance reveals that troublesome customers, defined as undecisive customers that keep cancelling their ubers after ordering them without the 5 minutes elapsing, are an increasing cost.

Your further discussion with the product engineering manager shows that there is an idea for creating a private rating of uber users based on this troublesome behaviour: users who cancel a large percentage of their trips will be given low ratings. And users with low ratings will not be 'actually assigned ubers' (even though the application may show otherwise) until a few minutes after they have ordered the uber.

a) Explain briefly how costs can be reduced with such a rating system

b) Suggest a refined question about saving costs, and what you expect to benefit from answering this question

c) What would be a way to answer this question with the given data

d) Suggest a hypothesis test, stating the null and alternate hypothesis. Assume here that if the user cancels 30 percent or more of their rides then they will get low ratings

e) Perform the test on the attached dataset, are you inclined to accept the null or alternate hypothesis explain your choice

f) Given the user data you analysed is representative of 1000 users, and assuming that cancellations within 5 minutes cost on average 3 SAR, how much money do you think you can save and over how many months?

Presentation

Problem 5:

Communicate your problem, question, refined question, statistical test results and overall conclusions from Problem 4 to your manager using the necessary visualisations. You should use your results from the prior problems to inspire or encourage your final argument 3

Attachment:- Data.rar