The purpose of this study was to find correlations related to the variable number of impressions, likes, subscribers, and comments on each learning video keyword search on YouTube. This research uses quantitative methods and experiments with secondary data sources. Exploratory Data Analysis in machine learning using several libraries in Python programming produces image visualizations that provide information related to the dataset that has been processed, such as boxplot graphs, histograms, line plots, and correlation graphs. Exploratory Data Analysis with machine learning that we have done finds results on boxplot graphs on five variables showing a whisker more elongated upwards which states positive data results. The difference in this histogram chart is in the duration variable. On the line plot graph, we find the keywords learning videos learning mathematics have the advantage of four variables and the keywords of accounting learning videos one variable.
Exploratory Data Analysis using the correlation head map in the seaborn library shows that the like and comment variables strongly correlate with a value of 1. Duration variables have a low and negative correlation with other variables. The subscribers variable has a high correlation with the like variable 0.95. Thus, several indicators need to be considered in making learning videos, such as content or content of innovative and creative learning videos, so that the number of likes and comments becomes high. The length of time in learning videos does not affect the number of likes, subscribers, and comments.