Node Project 3 | Deepnote

import pandas as pd import numpy as np from sklearn.ensemble import RandomForestClassifier from sklearn.neighbors import KNeighborsClassifier from sklearn.model_selection import train_test_split

jobs = pd.read_csv('NYC_Jobs.csv')

jobs.shape

jobs.head(5)

!pip install plotly import plotly import plotly.express as px

jobs_by_cat = jobs.groupby(by=['Agency']).agg('median').reset_index() jobs_by_cat.head()

fig = px.bar(jobs_by_cat, x='Agency', y='Salary Range From', color='Agency', color_discrete_sequence=plotly.colors.qualitative.Prism, title='NYC Jobs' ) fig

This plot shows the lower bound of the salary range for each NYC Agency. The agencies with the highest median salary are the Office of Management and Budget and Parks & Recreation. The agencies with the lowest median salaries are the Department of Probation and City Planning.

fig = px.scatter(jobs, x='Salary Range From', y='Salary Range To', opacity=0.3,hover_data=['Salary Range From','Salary Range To'], color_discrete_sequence=plotly.colors.sequential.BuPu_r, title='Correlation of Starting and Maximum Salaries') #fig.show('notebook') fig

A positive correlation is observed between starting salary ("Salary Range From") and Maximum Salary ("Salary Range To"). This trend aligns with what is expected; as the base salary increases for any given job of any given category so should the maximum salary for that job. As for outliers, there is one outlier point past 150k as the starting salary that has a 225k maximum salary.

#grouping by career level and then taking averages of the columns job_salary = jobs.groupby(by=['Career Level']).agg('mean').reset_index() #creating a bar graph to display average salary ranges across career levels fig = px.bar (job_salary, x='Career Level', y='Salary Range To', range_y = (0,250000), hover_name='Career Level', color = 'Salary Range To', title='Average Salary Range Across Career Levels', ) fig

This graph shows the average salary ranges across different career levels. Workers in executive positions had the highest average salaries while students had the lowest average salary.

##The model predicts the career level depending upon the "Salary Range To" variable, using classification. avgsalaries = jobs.groupby(by=['Career Level']).agg('mean') #gives means of each instance the career level appears avgsalaries['Salary Range To']

cols_to_drop = ['Salary Frequency','Full-Time/Part-Time indicator','Title Code No','Salary Range From', 'Level', 'Process Date', 'Job ID', '# Of Positions', 'Agency','Posting Type','Business Title', 'Civil Service Title', 'Title Classification', 'Job Category','Work Location','Division/Work Unit','Job Description', 'Minimum Qual Requirements', 'Preferred Skills','Additional Information','To Apply','Hours/Shift','Work Location 1', 'Recruitment Contact', 'Residency Requirement', 'Posting Date','Post Until','Posting Updated'] df_ml = jobs.drop(columns=cols_to_drop) df_ml=df_ml.dropna() #had to drop missing values here df_ml #condensed data frame

df_ml['Career Level'] = df_ml['Career Level'].map({'Entry-Level':0,'Executive':1, 'Experienced (non-manager)':2, 'Manager':3, 'Student' :4}) #the different levels the model will classify based on "Salary Range To" df_ml_cleaned = pd.get_dummies(df_ml,drop_first=True) df_ml_cleaned.dropna(axis='columns',how='all') #https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.dropna.html df_ml_cleaned.head()

X = df_ml_cleaned.drop(columns=['Career Level']) y = df_ml_cleaned['Career Level'] print(X) #X is the data frame with solely the "Salary Range To variable"; input for the model X y

from sklearn.model_selection import train_test_split X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.33, random_state=42)

from sklearn import tree clf = tree.DecisionTreeClassifier()

clf.fit(X_train, y_train)

predicted = clf.predict(X_test) actual = np.array(y_test) print('Look at first few predictions:') print('Predicted Career Level: ',predicted) #idk if this part is right but 2 and 3 are the column numbers. print('Actual Career Level: ',actual)

from sklearn.metrics import accuracy_score accuracy_score(predicted,actual)

!pip install graphviz==0.14.2

#Visualization of Decision Tree: from IPython.display import display, HTML estimator = clf.estimators_[1] from sklearn.tree import export_graphviz import graphviz from sklearn.tree import export_graphviz import graphviz from sklearn.tree import export_graphviz from IPython.display import SVG from graphviz import Source #!pip install graphviz #from google.colab import files graph = Source(export_graphviz(estimator,out_file=None,feature_names=X_train.columns,class_names=['Career Level'],filled=True,rounded=True)) graph # save to file png_bytes = graph.pipe(format='png') with open('tree.png', 'wb') as f: f.write(png_bytes) files.download('tree.png') # display in notebook SVG(graph.pipe(format='svg'))

Recommended on Deepnote

Stock Market Analysis

The 10 Best Ways to Create NumPy Arrays

Wide Residual Networks