This guide continues its series on “How to know when to use statistical test in research“, providing an overview of the key considerations and common types of statistical tests, helping researchers navigate the decision-making process effectively.
Principal component analysis (PCA)
Use this method when you aim to simplify a dataset by determining its principal components. For instance, you might want to examine stock prices to pinpoint the main sources of variation.
Discriminant analysis is a statistical technique used to classify observations into predefined groups. It does this by finding a set of linear combinations of the predictor variables that best separates the groups. The primary goal is to predict the group membership of new observations based on their characteristics.
Discriminant Analysis
Discriminant analysis is a method used to sort things into different groups. It looks at the data and finds ways to tell the groups apart. The main idea is to figure out which group a new item belongs to based on its features.
Use this when you want to guess which group something belongs to based on many pieces of information. Example: You want to guess if a customer will stop using your service based on how they use it.
Canonical correlation analysis
Canonical correlation analysis looks at how two sets of different variables relate to each other, with all measurements taken from the same person.
For example, let us say we have two groups of variables: one related to exercise and the other related to health. In the exercise group, we might look at how fast someone can climb stairs, how quickly they can run a certain distance, the amount of weight they can lift on a bench press and the number of push-ups they can do in a minute. In the health group, we could consider their blood pressure, cholesterol levels, glucose levels, and body mass index.
So, we measure both types of variables for each person and try to understand the relationships between exercise and health.
Bayesian inference Test
Use when: You want to change probabilities based on new information. For example, you want to change the chance of something being true based on new facts.
Bayesian regression
These methods help you create a model to show the relationships between variables. By using Bayesian methods, you can make predictions and learn more about how changes in one thing might influence another. It is a systematic way to incorporate prior knowledge and new data to improve your understanding over time.
First, you start with what you already know or believe about the variables. This is called your prior knowledge. Then, you collect new information or data. Bayesian methods combine your prior knowledge with this new data. They update your beliefs and make your model more accurate. Moreover, Bayesian methods are flexible. They can handle complex problems where relationships between variables are not straightforward. These methods provide a probabilistic approach, meaning they give you a range of possible outcomes and the likelihood of each outcome happening. This is helpful because it shows the uncertainty in your predictions.
In addition, as you gather more data, Bayesian methods allow you to refine your model continuously. This iterative process helps improve the reliability of your conclusions. Bayesian methods are useful in various fields like medicine, economics, and engineering because they provide a detailed and adaptive way to analyze relationships between variables.
Bayesian networks Test
Use when you want to understand how different things affect each other using Bayesian methods. Example: You want to see how genes and diseases are linked.
Decision trees
Use when you want to sort things using a tree-like model. Example: You want to guess if a customer will leave based on their usage patterns.
Random forests Test
Make use of this test when: You want to sort things using many decision trees. For example, you want to guess a disease based on symptoms.
Support vector machines (SVMs)
Use this test when you need to sort items into categories using a dividing line. For example, if you want to predict whether customers will leave or stay based on their usage habits, this method can help you make that distinction.
Cluster analysis
Use this test when: You want to group similar things based on their features. For example, you might want to divide customers into groups according to their shopping.
The academic Hive webpage offers top-notch tools for student and researcher success. We also offer Consultancy Services; book a session today
