What are Descriptive Statistics?
Descriptive statistics are summary statistics that describe features of the sampled data rather than inferring properties of the general population from the sample. Inferential statistics are used to analyze and predict the features of an entire population whereas descriptive stats help researchers better understand the properties of a particular sample.
This is crucial for fine-tuning a model since these stats show how the deep learning software processes data and how the algorithm can be improved.
How are Descriptive Statistics used?
These statistics can be univariate (one variable) or bi/multivariate (two or multiple variables). With univariate analysis, the distribution of a single variable is most often studied, in particular:
- Central tendency - Such as mean, median, and mode
Dispersion - The range, quartiles or other groupings of the dataset
Measures of spread - Such as variance and standard deviation
With bi or multivariate analysis, the relationships between the variables are added to the analysis. Some of the most important statistics for study include:
- Correlation – Examples include Pearson’s r if both variables are continuous, or Spearman's rho if one or both are not continuous
- Covariance - The scale that variables are measured on
Linear slope – Useful in regression analysis to reflect the relationship between variables.