From problem to insights

1. From problem to insights

In this video, we'll start planning out our analysis and discuss the important points we need to consider.

2. The data-driven process

There are generally five main steps that underpin every data-driven process. The first step is defining a problem statement, which will guide the rest of the process. This consists of the business problem and its breakdown into analytical questions. The second step is to collect and prepare the necessary data. We already have the survey data available, but we could collect additional data if necessary. For example, depending on the employee's job type and years of experience, we could add the average salary and include this in our analysis. The third step is to perform data analysis. The last two steps are about communicating the results and making it possible to take action on our newly gained insights. We'll mainly focus on step three in this video.

3. Things to consider

The most important things we need to consider when planning our analysis are as follows. First, we need to consider the difference between quantitative and qualitative variables. This influences which kinds of graphs, statistics, and analysis methods we can use. Most real-world datasets contain a mix of both types. Second, we also need to select the right type of analytics. This depends on the objectives of the analysis. The analytical questions we defined will be our main guide here. In the following slides, we'll look closer at both points.

4. Quantitative vs. qualitative variables

Quantitative variables describe something with numbers, which can be counted or measured. We can also perform mathematical operations like summing or multiplying with quantitative variables. We typically have a wider range of methods available for these variables, and they are easier to analyze. An example quantitative variable would be the distance from work to home. Qualitative variables, on the other hand, describe something with categories; they are things that can be observed. Although a bit harder to analyze, there are methods available specific to qualitative variables. These typically use ranking if there is an order between the categories or by counting the categories. An example qualitative variable would be the type of transport taken to go to the workplace.

5. Types of analytics

These are the four main types of analytics: descriptive, diagnostic, predictive, and prescriptive analytics. Each type of analytics has its own main focus. Descriptive analytics focuses on summarizing and visualizing the data, diagnostic analytics on the causes of events, predictive analytics on possible outcomes, and prescriptive analytics on the best course of action to take. We can make use of one type or multiple types. It all depends on the questions we want to answer.

6. Selecting the right type of analytics

To help you find the best suited type of analytics, it is best to look at your analytical questions, find the keywords and match these to the right type of analytics. For example, let's look at one of the questions we defined earlier and its subquestions. In our case, the keywords are "distinguish" and "profiles". In other words, we want to find out whether there are different groups in the data. The subquestions refer to the possible characteristics of these groups. These questions best fit with summarizing and visualizing the data; therefore, descriptive analytics would be the most suited.

7. Let's practice!

Enough theory, time to dive into the case study again!

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.