RSSI Data Preparation for Machine Learning

In this study, we prepared unstructured raw RSSI (Received Signal Strength Indication) data to prepare it for use in machine learning predictive modeling. RSSI data preparation is very important in predictive modeling machine learning projects, such as classification and regression, so that the algorithm can model it based on the right conditions, because the data has a more specific part but does not necessarily affect the algorithm performance. RSSI data were collected from the Dinamika Bangsa University Campus building, namely 225 reference points. The two main processes involved in data compilation, namely construction data and data cleaning. From the process of identifying a column containing a single RSSI value, several unique values were found in each column in the RSSI data table. Since no column has a Single Value, no columns are deleted. Then in the process of deleting columns that have low RSSI variance, the input is 5969 rows and 18 columns. The output is 5969 rows and 3 columns. However, the number of columns that passed the Low Variance selection was 5969 rows and 18 columns, meaning that all columns in the dataset have high variance. Additionally, no duplicate RSSI data were found. It can be said that the result of the overall preparation of this data set is the RSSI data used in good condition and can be used for machine learning needs.