Subsets according time windows for Coffee Leaf Rust Incidence modeling. Paper: Discovering weather periods and crop properties favorable for coffee rust incidence from feature selection approaches

The climate dataset was processed to generate four data subsets corresponding to four time windows: 3, 4, 7, and 14 consecutive days. We use the concept of time windows to generate consecutive subperiods of each climate variable within the main period of 14 days before the date of prediction. (DP). This process generates new attributes corresponding to each variable. The number of periods depends on the size of the window e.g., the window of 4 consecutive days generates 11 new sub-periods for each climatic variable. The index that characterizes it indicates the days covered by the window e.g., tMin11-8 corresponds to the minimum temperature between days 11 and 8 before DP. We called the generated subsets 3D, 4D, 7D, 14D. Each subset had 439 instances, and the dimension depended on window size: 14D had 13 variables (8 related to climate), 7D had 69 features (64 related to climate), 4D had 93 features (88 related to climate), and 3D had 101 features (96 related to climate). The target variable was predicted Coffee Leaf Rust Incidence (pCLRI), and the predictors were the rest of the experiment variables: current CLRI (cCLRI), shade, host growth (hGrowt), management (mgmt) and climatic variables: maximum (tMax) and minimum (tMin) air temperature, average (tAvg) air temperature calculated over the day, average (hAvg) and minimum (hMin) relative humidity, daily precipitation (pre). The data in the files did not contain null data. The thermal amplitude (tAmp), which represents the difference between the maximum and minimum temperatures, and the characterization of each day as a rainy day or not (precipitation greater or equal to 1 mm) (rDay)本气候数据集经处理后生成了四个分别对应连续3天、4天、7天和14天时间窗口的数据子集。我们采用时间窗口的概念，在预测日期（DP）前的14天主周期内，为每个气候变量生成连续子时段。该过程会为每个变量生成新属性，子时段的数量取决于窗口大小：例如，连续4天的窗口会为每个气候变量生成11个新子时段。用于表征该窗口的索引可指明窗口覆盖的日期范围，例如tMin11-8代表预测日期（DP）前第11天至第8天的最低气温。我们将生成的子集命名为3D、4D、7D、14D。每个子集均包含439条实例，特征维度随窗口大小变化：14D子集包含13个变量（其中8个与气候相关），7D子集包含69个特征（其中64个与气候相关），4D子集包含93个特征（其中88个与气候相关），3D子集包含101个特征（其中96个与气候相关）。本数据集的目标变量为预测性咖啡叶锈病发病率（predicted Coffee Leaf Rust Incidence，pCLRI），预测因子为实验中其余变量：当前咖啡叶锈病发病率（current CLRI，cCLRI）、遮荫率、寄主生长量（hGrowt）、管理措施（mgmt）以及气候变量：最高气温（tMax）、最低气温（tMin）、日平均气温（tAvg）、平均相对湿度（hAvg）、最低相对湿度（hMin）与日降水量（pre）。数据文件中不存在空值。此外还计算了代表最高与最低气温差值的气温日较差（tAmp），以及表征当日是否为降雨日的变量（日降水量≥1 mm时记为降雨日，rDay）