使用LSTM进行时间序列问题形成输入数据

发布于 2025-01-19 09:34:18 字数 1010 浏览 2 评论 0原文

我有一个数据集，其中包含有关植物的信息，例如 ID、年龄、重量、花瓣颜色、高度、获取信息的日期和时间。身高是目标，所以我想预测身高。

我的训练数据集示例：

|Date | time | ID | color | weight(kilogram) | height(centimeter) | age ( in days) | 
    | ----| -----| -- | ---- -| ---------------- | -------------------| ---------------| 
      15/09/2021| 8pm |14 | yelow | 2 |50|12
      13/04/2021| 8am |3 | pink| 1 |25|10
      13/04/2021| 10pm |3| pink| 1 |30|10
      4/06/2021 | 4pm |5 | green| 2 |50|5

我们可以获得一天中不同时间的植物信息。我想根据几个特征来预测大小。

我的测试数据集示例：

 |Date | time | ID | color | weight(kilogram) | age ( in days) | 
    | ----| -----| -- | ---- -| ---------------- | ---------------| 
      15/09/2021| 8pm |15| yelow | 2 |12
      13/04/2021| 8am |13 | pink| 1 |10
      13/04/2021| 10pm |13| pink| 1 |10
      4/06/2021 | 4pm |55 | green| 2 |5

在预测期间，我们具有与训练中相同的特征，但没有高度（目标）。例如，我只想用我的特征进行预测，而不是在一周内进行完整的预测。

我想考虑暂时性的概念，即根据过去的高度来预测高度。

我很难理解输入表格。有人可以向我解释带有这种类型的数据集的 lstm 中输入数据的形式吗？

先感谢您

原文

I have a dataset which contains informations about plants like ID,age, weight, petal color, height, date and time on which the information is taken. Height is the target, so i want to predict height.

Example of my train dataset:

|Date | time | ID | color | weight(kilogram) | height(centimeter) | age ( in days) | 
    | ----| -----| -- | ---- -| ---------------- | -------------------| ---------------| 
      15/09/2021| 8pm |14 | yelow | 2 |50|12
      13/04/2021| 8am |3 | pink| 1 |25|10
      13/04/2021| 10pm |3| pink| 1 |30|10
      4/06/2021 | 4pm |5 | green| 2 |50|5

We can have information on plants at different times of the day. I would like to predict the size based on several features.

Example of my test dataset:

 |Date | time | ID | color | weight(kilogram) | age ( in days) | 
    | ----| -----| -- | ---- -| ---------------- | ---------------| 
      15/09/2021| 8pm |15| yelow | 2 |12
      13/04/2021| 8am |13 | pink| 1 |10
      13/04/2021| 10pm |13| pink| 1 |10
      4/06/2021 | 4pm |55 | green| 2 |5

During the prediction we have the same features as in the training without the height (target). I only want to make predictions with my features and not make a complete prediction over a week for example.

I would like to take into account the notion of temporality, i.e. make a prediction of the height according to the past height.

I am having difficulty understanding the input form. Can someone explain to me the form of input data in an lstm with with a dataset of this type?

thank you in advance

分享到QQ

分享到微博