Type Conversion

Why should we care about type?

• Reduce memory usage: Generally speaking, the numerical value would be regarded as float64 or int64. In most cases, this is OK. However, imagine you have, say, 50 million rows, but the columns will only store numbers from 0 to 20. int8 is quite enough in this case and saves a lot of memory.
• Unmatched type: As mentioned above, the numerical value would be regarded as float64 or int64. In most cases, it’s ok. However, if one column is an ID column, obviously, float64 is not a suitable type.
• Specifical type: Datetime is one special case. If you don’t specify it, the Datetime would be regarded as an object type.