Explanation: This is a machine learning practical project (with data + code + documentation + video explanation). If you need data + code + documentation + video explanation, you can go directly to the end of the article to get it.
1. Project Background
PSO is the English abbreviation of Particle Swarm Optimization (Particle Swarm Optimization), which is a population-based stochastic optimization technique proposed by Eberhart and Kennedy in 1995. Particle swarm algorithm imitates the flocking behavior of insects, herds of animals, flocks of birds and fish, etc. These groups look for food in a cooperative way, and each member of the group constantly changes by learning its own experience and the experience of other members. its search mode.
This project uses PSO particle swarm optimization BP neural network algorithm to build a regression model.
2. Data Acquisition
The modeling data for this time comes from the Internet (compiled by the author of this project), and the statistics of the data items are as follows:
The data details are as follows (partial display):
3. Data preprocessing
3.1 View data with Pandas tool
Use the head() method of the Pandas tool to view the first five rows of data:
key code:
3.2 Data missing view
Use the info() method of the Pandas tool to view data information:
As can be seen from the figure above, there are a total of 11 variables, no missing values in the data, and a total of 2000 data.
key code:
3.3 Data descriptive statistics
Use the describe() method of the Pandas tool to view the mean, standard deviation, minimum, quantile, and maximum of the data.
The key code is as follows:
4. Exploratory Data Analysis
4.1 Histogram of y variables
Use the hist() method of the Matplotlib tool to draw a histogram:
As can be seen from the figure above, the y variable is mainly concentrated between -400 and 400.
4.2 Correlation analysis
As can be seen from the figure above, the larger the value, the stronger the correlation. A positive value is a positive correlation, and a negative value is a negative correlation.
5. Feature engineering
5.1 Establish feature data and label data
The key code is as follows:
5.2 Dataset splitting
Use the train_test_split() method to divide according to 80% training set and 20% test set. The key code is as follows:
6. Construct PSO particle swarm optimization BP neural network regression model
Mainly use the PSO particle swarm optimization algorithm to optimize the BP neural network regression algorithm for target regression.
6.1 PSO particle swarm optimization algorithm to find the optimal parameter value
Iterate process data:
Optimal parameters:
6.2 Optimal parameter value construction model
6.3 Summary information of the optimal model
As can be seen from the figure above, the model has a total of 1441 parameters, as well as the parameters of each layer.
6.4 Structure of the Optimal Model
As can be seen from the above figure, the input and output of each layer and the result correlation relationship between layers.
6.5 Model Loss Visualization
As can be seen from the above figure, the loss of the training set and the test set gradually decreases with the increase of the number of iterations, and the loss gradually stabilizes after 10 iterations.
7. Model evaluation
7.1 Evaluation indicators and results
The evaluation indicators mainly include explainable variance value, mean absolute error, mean square error, R square value and so on.
It can be seen from the above table that the R square is 0.9941, which means that the model has a better effect.
The key code is as follows:
7.2 Comparison chart of actual value and predicted value
From the above figure, it can be seen that the fluctuations of the actual value and the predicted value are basically the same, and the model fitting effect is good.
8. Conclusion and Outlook
In summary, this project uses the PSO particle swarm algorithm to find the optimal parameter value of the BP neural network algorithm to build a regression model, which finally proves that the model we proposed works well. This model can be used for forecasting of everyday products.
# y变量分布直方图
fig = plt.figure(figsize=(8, 5)) # 设置画布大小
plt.rcParams['font.sans-serif'] = 'SimHei' # 设置中文显示
plt.rcParams['axes.unicode_minus'] = False # 解决保存图像是负号'-'显示为方块的问题
data_tmp = df['y'] # 过滤出y变量的样本
# 绘制直方图 bins：控制直方图中的区间个数 auto为自动填充个数 color：指定柱子的填充色
plt.hist(data_tmp, bins='auto', color='g')
plt.xlabel('y') # 设置x轴名称
plt.ylabel('数量') # 设置y轴名称
plt.title('y变量分布直方图') # 设置标题的名称
plt.show() # 显示图片
# ******************************************************************************
# 本次机器学习项目实战所需的资料，项目资源如下：
# 项目说明：
# 链接：https://pan.baidu.com/s/1c6mQ_1YaDINFEttQymp2UQ
# 提取码：thgk
# ******************************************************************************
if abs(params[0]) > 0: # 判断取值
units = int(abs(params[0])) + 10 # 赋值
else:
units = int(abs(params[0])) + 16 # 赋值
if abs(params[1]) > 0: # 判断取值
epochs = int(abs(params[1])) + 10 # 赋值
else:
epochs = int(abs(params[1])) + 20 # 赋值
For more project practice, see the list of machine learning project practice collections:
List of actual combat collections of machine learning projects