I will inspect the data and check for null values, duplicates, erroneous data types or other incoherences in the data, then I will take the proper action be it removal of corrupt data, reformatting it, or filling it with statistically adequate values. I will also assess the need for certain attributes to remain in the dataset or not and I will create, whenever necessary and beneficial, new engineered features and values to have a dataset ready for exploration and analysis.
I can use a few programs to do that starting with Excel, Power BI, SQL, SAS, or Python.