1. Choosing the best analytical solution
Welcome. In this video, we will cover best practices to help us identify the most appropriate analytical solution. An analytical solution that answers the analytical question and solves the business problem.
2. Steps to identify the best analytical technique
To make sure the best analytical solution is selected, we can follow these three high level steps.
First, categorize the analytical question into the relevant type of analytical solution.
Then, brainstorm relevant techniques related to the question and the type of analytical solution.
Finally, select the most appropriate technique to implement.
Let’s go through these steps in more detail.
3. Find the relevant type of analytics
A good practice before choosing the analytical solution is to categorize the analytical question into one of the four types of analytical solutions: prescriptive, diagnostic, predictive, or prescriptive analytics. Each type of analytical solution requires different analytical techniques and algorithms. By categorizing the analytical question, it helps to narrow down the techniques available and makes it easier to select the appropriate technique or algorithm for the analysis.
4. Consider the output of the analysis
Once the type of analytics solution is known then the next step is to brainstorm relevant techniques under the respective type. During this process it is important to consider the following:
Think of what the desired output should be. This ensures that the solution can address the analytical question. For example, if we want to predict which customer segments are most likely to churn then we should select a predictive analytics technique that can provide a binary prediction as output, churn, or no churn. On the other hand, If we want to summarize time series data such as sales over time, we should use a descriptive analytics technique to easily communicate the insights. This could be through data visualization and line charts.
5. Consider the data type and size
The type of data that will be used in the analysis is another important factor to consider. Depending on whether the data is continuous or categorical, structured or unstructured, different methods or techniques may be more appropriate. For example, analyzing text data such as customer reviews would require different techniques compared to numerical data.
The size of your sample can also impact the analytical technique selected. If we have a large sample, we might be able to use more complex techniques, such as machine learning algorithms, whereas, with a smaller sample, we might need to use simpler techniques.
6. Consider time and skill set
When selecting the final analytical technique it is important to consider the time and resources of the analytics team. If time is limited and the solution or insights are needed fast by the business to make decisions, then simpler techniques that can be implemented quickly may be preferred over more complex techniques that require more time to implement.
In addition, the skill set of the analytics team can also impact the choice of the analytical technique. Some techniques such as machine learning or optimization may require expertise that the team might not possess, while other techniques may be more familiar and accessible.
7. Aim for the simplest solution
The complexity of an analytical solution does not necessarily translate into better results. In many cases, a simple solution may be more effective than a complex one.
The objective of any analytical solution is to provide useful insights and solve the problem at hand. If a simple solution can provide the required insights, there may be no need for a more complex solution. Complex analytical solutions can also be challenging to implement and may require specialized expertise, which can increase the cost and time required to implement the solution. In addition, complex analytical solutions can sometimes be challenging to interpret and communicate to non-technical stakeholders. In contrast, simpler solutions can be more easily understood and communicated to a broader audience. Therefore, the best analytical solution is one that is fit for purpose and can provide actionable insights that can drive decision-making effectively.
8. Let's practice!
Now, let's put that knowledge into practice!