Classification and prediction of heartbeat signal

Task1. Understanding of competition questions

Time of the game ：3 month 12 Japan ——5 month 12 Japan

The competition questions are based on ECG heartbeat signal data , Ask contestants Predict the category of heartbeat signal according to ECG induction data , Among them, the heartbeat signal corresponds to normal cases and cases affected by different arrhythmias and myocardial infarction , This is a Many classification The problem of .

1. Learning goals ： Understand the competition data and scoring system
2. Learning tasks ： Download and understand the game data
3. Understanding of competition questions ：
   1. According to the data set provided by the competition , Choose to build the appropriate model , Predict different types of heartbeat signals . The task of the competition is to predict the type of ECG signal , The total amount of competition data exceeds 20 ten thousand , Extract from it 10 Ten thousand as a training set ,2 Ten thousand as a test set A,2 Ten thousand as a test set B.
   2. In the training data ,id The unique ID assigned to the heartbeat signal ,artbeat_signals Heartbeat signal sequence ( Between the data “,” separation ),label Heartbeat signal category （0、1、2、3）. In the test data ,id Unique identification of heartbeat assignment heartbeat_signals Heartbeat signal sequence ( Between the data “,” separation ).
   3. Find the absolute values of four predicted signals and real signals , The smaller the absolute value, the better .
4. Game Analysis ：
   1. This competition question is a multi classification problem , There are four categories .
   2. Main application xgb、lgb、catboost, as well as pandas、numpy、matplotlib、seabon、sklearn、keras And so on data mining common library or framework for data mining tasks .