Microsoft BI or Tableau: recommendation on DB design for an existing key-value pair db

Since freelancer is removing line breaks, you might look at the attached text file with a better layout. I have a table that contains about 5,000,000 records. It is a key-value pair and It has recordID, field_name, field_value, date_created. There are 3,000 different field_names. Some values have JSON with some additional properties. Ex. recordID; field_name; field_value 1; name; John; 2; name; Steve; 1; reg_date; Oct 1, 2018 1; age; 21 1; car; ("carType": ["Truck", "SUV"]) 2; car; ("carType": ["Sedan"]) 3; some_field; ("gameType": ["6","5","7","1","9","3","4","6","66","88","66","2","123"]) 1; skin_color; Blue 1; skin_color; Green 1; teeth; 6 1; skin_color; Yellow 1; domestic, Yes In this case, for recordID 1, it is John, 21 year old, has a truck and SUV, with 6 teeth, and the colors are Blue, Green, Yellow. Steve drives a sedan. Table 4 contains 5MM records (about 5,000 different gamers, but each of them might have a completely different set of field_names, on average each gamer has about 1,000 of them) I would like to know what would be the best way to change this table so that I can use Microsoft BI or Tableau graphical interface and for example select a pie chart representing all the gamers where x-axis = Registration Date, y-axis = number of gamers, and I can filter by Gamer's Age and Skin Color. Please remember: there are 3,000 different variables (field_names) I could be reporting on. They are not really categorized, they are just various random properties. I don't think it's a best practice to have a big table with about 3,000 columns and then creating a separate table that splits JSON contents. Again, the end goal is to use a visual GUI with friendly field names and create various graphs with filters. What would be your reco on changing the DB architecture to quickly analyze the info using friendly names in Microsoft BI or Tableau?