In the real-life dataset, Datetime and coordinates features are often present. To get useful information from Datetime and Coordinates features are also a challenging task. Here, I have explained to extract features from Datetime and Coordinates features with example in Python.
Datetime:
Qbay e-commerce shopping brand want to predict future sales. The Data contain previous 4 years records(2015-2019) of sales on each day. Here, Date is a very important feature and play a significant role to make better predictions. Let’s find some beautiful features from the date feature.
Pandas is a very beautiful python package, which helps to get a month, year, day etc. features from date.
data['timestemp'] = pd.to_datetime(data['timestemp'], format = '%Y/%m/%d %H:%M:%S') timestemp_dt= data['timestemp'].dt
data['day'] = timestemp_dt.day data['day_name'] = timestemp_dt.day_name data['week'] = timestemp_dt.week # A week ordinal of the year data['weekOfyear'] = timestemp_dt.weekofyear data['dayofweek'] = timestemp_dt.dayofweek data['dayofyear'] = timestemp_dt.dayofyear data['quarter'] = timestemp_dt.quarter data['month'] = timestemp_dt.month data['month_name'] = timestemp_dt.month_name
data['is_month_start'] = timestemp_dt.is_month_start # Indicates whether the date is the first day of the month. data['is_month_end'] = timestemp_dt.is_month_end # Indicates whether the date is the last day of the month. data['is_quarter_start'] = timestemp_dt.is_quarter_start data['is_quarter_end'] = timestemp_dt.is_quarter_end data['is_year_start'] = timestemp_dt.is_year_start data['is_year_end'] = timestemp_dt.is_year_end data['is_leap_year'] = timestemp_dt.is_leap_year data['daysinmonth'] = timestemp_dt.daysinmonth # The number of days in the month.
data['timezone'] = timestemp_dt.tz # Return timezone, if any. data['hour'] = timestemp_dt.hour data['minute'] = timestemp_dt.minute data['second'] = timestemp_dt.second
Coordinates:
Coordinates or location features are most often seen in real-life datasets. We can extract helpful information from coordinates and location features like an area are reach or poor, distance to the nearest hospital, school or shopping centre from particular coordinates or locations. We can extract information from existing data and also using additional data from the internet.
For Example, A house price prediction data contain feature neighbourhood location of the house.
Get richest and poorest area Is_metrocity Get distance to nearest hospital Get distance to nearest School Get distance to nearest Shopping area Get distance to airport
We can also extract weather information for a particular location
Max/Min temperature in winter Max/Min temperature in summer Max/min temperature in monsoon