Explore the data
Data Samples
Test Hypothesis
Data Modeling
Scatter Plot for correlation between clicks and conversions
Inu + Neko had a lot of success with a marketing campaign and is seeing a lot of growth. The company wants to know how there marketing strategy and how to improve it.
When you open the Capstone Dataset spreadsheet you can see that it is a collection of data comparing the performance of two separate ad campaigns: a Facebook Ad campaign (in the blue-colored columns of the sheet) and an AdWords Ad campaign (in the green-colored columns). The data is collected for every day of the year 2019, so you have 365 lines of campaign data to analyze! Ultimately, we want to determine whether one of these ad campaigns is more effective, in terms of conversions, compared to the other. But before we address that question, we should get a better understanding of the conversion data for each of the campaigns.
To better understand how effective our campaigns are at producing conversions, we’re going to begin by analyzing the relationship between a campaign’s number of clicks and the corresponding number of conversions.
we can do this by calculating
(1.) the measures of central tendency,
(2.) some measures of dispersion, and
(3.) the frequency and correlation data for a campaign’s clicks and conversions.
1. Histograms for ‘Clicks’ and ‘Conversions’ Data
2. Determine What Variable Types You’re Working With
In this part, we have some important questions to ask about our data samples. The answers to these questions can be essential to determine what kind of analysis we will ultimately be able to perform with data! For example, histograms for our data samples, which help determine what kind of shape best describes the distribution of your data. It’s important to know this because many analyses require that your data has a specific kind of distribution shape (for example, a normal distribution curve).
We will use experimental design, and marketing studies, in particular, to formulate and test a hypothesis
And we will ask one specific question which is
"Is there a difference between the number of conversions on the Facebook platform versus the AdWords platform?"
We used a full design process involving five basic steps:
Questioning
Hypothesis
Required Variables
Choosing a Measurement Approach
Selecting an Analysis
In this part we try to answer these questions:
what kind of conversion rates I can expect from my Facebook Ads. But which model to run for analysis depends on the question we are trying to answer and the type of data have to work with. So, for this part, we want to go through the process of choosing a model that’s appropriate to the evaluation question and the Dataset that have been working with. Specifically, now that we’ve determined (in 3rd part) that Facebook Ad Clicks are most effective at generating Conversions, let’s get a clearer handle on this effectiveness by answering this question:
"How many Facebook ad conversions do I expect given a certain number of Facebook ad clicks?"