文章目录
1. 读取数据2. 处理label3. 添加特征4. 数据集切片5. 训练6. 预测learn from /learn/feature-engineering
下一篇：Feature Engineering 特征工程 2. Categorical Encodings
1. 读取数据
预测任务：用户是否会下载APP，当其点击广告以后
数据集：ks-projects-01.csv
读取数据，指定两个特征'deadline','launched'，parse_dates解析为时间
ks = pd.read_csv('ks-projects-01.csv',parse_dates=['deadline','launched'])
预测Kickstarter项目是否会成功。state作为结果label
可以使用类别category，货币currency，资金目标funding goal，国家country以及启动时间launched等特征
2. 处理label
准备标签列，看看有哪些值，转换成可用的数字格式pd.unique(ks.state)
有6种数值
array(['failed', 'canceled', 'successful', 'live', 'undefined','suspended'], dtype=object)
每种多少个？按state分组，每组中ID行数有多少
ks.groupby('state')['ID'].count()
statecanceled 38779failed 197719live 2799successful 133956suspended 1846undefined 3562Name: ID, dtype: int64
简单处理下标签列，正在进行的项目live丢弃，successful的标记为1，其余的为0
ks = ks.query('state != "live"') # live行不要ks = ks.assign(outcome=(ks['state']=='successful').astype(int))# label 转成1,0，int型
3. 添加特征
把launched时间拆分成，年月日小时，作为新的特征ks = ks.assign(hour=ks.launched.dt.hour,day=ks.launched.dt.day,month=ks.launched.dt.month,year=ks.launched.dt.year)ks.head()
转换文字特征category, currency, country为数字
from sklearn.preprocessing import LabelEncodercat_features = ['category','currency','country']encoder = LabelEncoder()encoded = ks[cat_features].apply(encoder.fit_transform)encoded.head(10)
将选择使用的特征合并在一个数据里
X = ks[['goal', 'hour', 'day', 'month', 'year', 'outcome']].join(encoded)X.head()
4. 数据集切片
数据切片，按比例分成训练集、验证集、测试集（0.8，0.1，0.1）更高级的简单做法sklearn.model_selection.StratifiedShuffleSplitvalid_ratio = 0.1valid_size = int(len(X)*valid_ratio)train = X[ : -2*valid_size]valid = X[-2*valid_size : -valid_size]test = X[-valid_size : ]
需要关注下，label 在每个数据集中的占比是否接近
for each in [train, valid, test]:print("Outcome fraction = {:.4f}".format(each.outcome.mean()))
Outcome fraction = 0.3570Outcome fraction = 0.3539Outcome fraction = 0.3542
5. 训练
使用LightGBM模型进行训练机器学习算法之LightGBM
feature_cols = train.columns.drop('outcome')dtrain = lgb.Dataset(train[feature_cols], label=train['outcome'])dvalid = lgb.Dataset(valid[feature_cols], label=valid['outcome'])param = {'num_leaves': 64, 'objective': 'binary'}param['metric'] = 'auc'num_round = 1000bst = lgb.train(param, dtrain, num_round, valid_sets=[dvalid],early_stopping_rounds=10, verbose_eval=False)
6. 预测
对测试集进行预测from sklearn import metricsypred = bst.predict(test[feature_cols])score = metrics.roc_auc_score(test['outcome'], ypred)print(f"Test AUC score: {score}")
下一篇：Feature Engineering 特征工程 2. Categorical Encodings
