Job Description
Content of a given training set PM2.5 train.csv, prediction requires monitoring of the air in the first nine hours of 10 hours.
Training set Description:
(1): CSV file, meteorological data including Taiwan Fengyuan area 240 days (20 days before taking a month to do the training data set, December X20 days = 240 days, 10 days after the monthly data for testing, students are not visible);
(2): daily monitoring time is 0, ...... 23 when the total time of the node 1 24;
(3): the day indicator comprises a gas concentration detecting CO, NO, PM2.5, PM10 and the like, whether rain, wind and other weather information, a total of 18;
(4): data set https://github.com/datawhalechina/leeml-notes/blob/master/docs/Homework/HW_1/Dataset
data processing
[ "Data frame" mentioned below refers to the data structure does not DataFrame pandas library, but rather refers to a two-dimensional data packet]
The apparent operational requirements, need to use the weather observation data for nine consecutive time points to predict time PM2.5 content of 10 points. For it every day, the information it contains is the dimension (18, 24) (18 indicators, 24 time node). Data 0-8 may be truncated at
Taken out, is formed in one dimension (18,9) data frame, as training data, the content of the PM2.5 9 is taken out, as the training data corresponding label; 1-9 Similarly desirable when data as the training with the data frame, PM2.5 label content 10 as this ......
Segmentation, day information may be divided into shape of 15 (18,9) and the corresponding data frame 15 label.