```python
import pandas as pd
import matplotlib.pyplot as plt
```

```python
data = pd.read_csv("upsampling_whole_sep.csv", index_col=0)
data["MSE / 10"] = data["MSE"] / 10
data.head()
```

```python
data[data.type == "NB"][["MSE / 10", "acc", "bal acc", "upsampling"]].plot(kind="bar", x="upsampling")
plt.xticks(rotation=45)
plt.xlabel('')
plt.ylabel("Score on Naive Bayes")
plt.title("Different upsampling method with Naive Bayes")

plt.savefig("figs/NB_upsampling.jpg", bbox_inches='tight')
plt.show()
```

```python
data[data.type == "Tree"][["MSE / 10", "acc", "bal acc", "upsampling"]].plot(kind="bar", x="upsampling")
plt.xticks(rotation=45)
plt.xlabel('')
plt.ylabel("Score on Regression Tree")
plt.title("Different upsampling method with Regression Tree")

plt.savefig("figs/Tree_upsampling.jpg", bbox_inches='tight')
plt.show()
```

```python
data[data.type == "LR"][["MSE / 10", "acc", "bal acc", "upsampling"]].plot(kind="bar", x="upsampling")
plt.xticks(rotation=45)
plt.xlabel('')
plt.ylabel("Score on Ridge Regression")
plt.title("Different upsampling method with Ridge Regression")

plt.savefig("figs/LR_upsampling.jpg", bbox_inches='tight')
plt.show()
```

```python
data[data.type == "SVM"][["MSE / 10", "acc", "bal acc", "upsampling"]].plot(kind="bar", x="upsampling")
plt.xticks(rotation=45)
plt.xlabel('')
plt.ylabel("Score on SVM")
plt.title("Different upsampling method with SVM")

plt.savefig("figs/SVM_upsampling.jpg", bbox_inches='tight')
plt.show()
```

```python
data[data.type == "NN"][["MSE / 10", "acc", "bal acc", "upsampling"]].plot(kind="bar", x="upsampling")
plt.xticks(rotation=45)
plt.xlabel('')
plt.ylabel("Score on Neural Network")
plt.title("Different upsampling method with Neural Network")

plt.savefig("figs/NN_upsampling.jpg", bbox_inches='tight')
plt.show()
```

```python
data_no_border = pd.read_csv("upsampling_noborder_sep.csv", index_col=0)
data_no_border["MSE / 10"] = data_no_border["MSE"] / 10
data_no_border.head()
```

```python
assert (data_no_border["type"] == data["type"]).all() and (data_no_border["upsampling"] == data["upsampling"]).all()

data["MSE trimmed / 10"] = data_no_border["MSE / 10"]
data["acc trimmed"] = data_no_border["acc"]
data["bal acc trimmed"] = data_no_border["bal acc"]
data.head()
```

```python
data[data.type == "NB"][["MSE / 10", "MSE trimmed / 10", "acc", "acc trimmed",  "bal acc", "bal acc trimmed", "upsampling"]].plot(kind="bar", x="upsampling")
plt.xticks(rotation=45)
plt.xlabel("")
plt.ylabel("Score on Naive Bayes")
plt.title("Inflence of trimming to upsampling method with Naive Bayes")

plt.savefig("figs/trimmed_NB_upsampling.jpg", bbox_inches='tight')
plt.show()
```

```python
data[data.type == "Tree"][["MSE / 10", "MSE trimmed / 10", "acc", "acc trimmed",  "bal acc", "bal acc trimmed", "upsampling"]].plot(kind="bar", x="upsampling")
plt.xticks(rotation=45)
plt.xlabel("")
plt.ylabel("Score on Regression Tree")
plt.title("Inflence of trimming to upsampling method with Regression Tree")

plt.savefig("figs/trimmed_Tree_upsampling.jpg", bbox_inches='tight')
plt.show()
```

```python
data[data.type == "LR"][["MSE / 10", "MSE trimmed / 10", "acc", "acc trimmed",  "bal acc", "bal acc trimmed", "upsampling"]].plot(kind="bar", x="upsampling")
plt.xticks(rotation=45)
plt.xlabel("")
plt.ylabel("Score on Ridge Regression")
plt.title("Inflence of trimming to upsampling method with Ridge Regression")

plt.savefig("figs/trimmed_LR_upsampling.jpg", bbox_inches='tight')
plt.show()
```

```python
data[data.type == "SVM"][["MSE / 10", "MSE trimmed / 10", "acc", "acc trimmed",  "bal acc", "bal acc trimmed", "upsampling"]].plot(kind="bar", x="upsampling")
plt.xticks(rotation=45)
plt.xlabel("")
plt.ylabel("Score on SVM")
plt.title("Inflence of trimming to upsampling method with SVM")

plt.savefig("figs/trimmed_SVM_upsampling.jpg", bbox_inches='tight')
plt.show()
```

```python
data[data.type == "NN"][["MSE / 10", "MSE trimmed / 10", "acc", "acc trimmed",  "bal acc", "bal acc trimmed", "upsampling"]].plot(kind="bar", x="upsampling")
plt.xticks(rotation=45)
plt.xlabel("")
plt.ylabel("Score on Neural Network")
plt.title("Inflence of trimming to upsampling method with Neural Network")

plt.savefig("figs/trimmed_NN_upsampling.jpg", bbox_inches='tight')
plt.show()
```

```python
data = data.rename(columns={"bal acc": "balAcc"})
data["upsampling"] = data["upsampling"].apply(lambda x: "no upsampling" if x == "no_upsampling" else x) 

data["MSE"] = data["MSE"].apply(lambda x: round(x, 4))
data["acc"] = data["acc"].apply(lambda x: round(x, 4))
data["balAcc"] = data["balAcc"].apply(lambda x: round(x, 4))
(data[data["type"] == "NN"][["MSE", "acc", "balAcc", "upsampling"]]).to_csv("figs/NN_upsampling.csv", index=False)
```

```python
data_no_border = data_no_border.rename(columns={"bal acc": "balAcc"})
data_no_border["upsampling"] = data_no_border["upsampling"].apply(lambda x: "no upsampling" if x == "no_upsampling" else x) 

data_no_border["MSE"] = data_no_border["MSE"].apply(lambda x: round(x, 4))
data_no_border["acc"] = data_no_border["acc"].apply(lambda x: round(x, 4))
data_no_border["balAcc"] = data_no_border["balAcc"].apply(lambda x: round(x, 4))
(data_no_border[data_no_border["type"] == "NN"][["MSE", "acc", "balAcc", "upsampling"]]).to_csv("figs/NN_trimmed_upsampling.csv", index=False)
```

```python

```