
%matplotlib inline
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
import os, datetime, warnings
print('Last updated on {}'.format(datetime.datetime.now().ctime()))

Last updated on Wed Feb  1 22:09:59 2023


import sns_settings
sns.set_context('notebook')
pd.options.display.max_columns = 50
warnings.filterwarnings('ignore')


df_test_dic1 = pd.read_csv('results/bottle_data_test_dic_all_vars.csv', index_col=0)

df_work = pd.melt(df_test_dic1, value_vars=df_test_dic1.columns.tolist()[:8], id_vars=['DIC residuals'])
g = sns.relplot(x='value', y='DIC residuals', data=df_work, col='variable', col_wrap=4, facet_kws={'sharex':False})
_ = g.set_titles(template='{col_name}', row_template='{col_name}')


df_test_dic2 = pd.read_csv('results/bottle_data_test_dic_noO2.csv', index_col=0)

df_work = pd.melt(df_test_dic2, value_vars=df_test_dic2.columns.tolist()[:7], id_vars=['DIC residuals'])
g = sns.relplot(x='value', y='DIC residuals', data=df_work, col='variable', col_wrap=4, facet_kws={'sharex':False})
_ = g.set_titles(template='{col_name}', row_template='{col_name}')


df_test_dic3 = pd.read_csv('results/bottle_data_test_dic_nosat.csv', index_col=0)

df_work = pd.melt(df_test_dic3, value_vars=df_test_dic3.columns.tolist()[:4], id_vars=['DIC residuals'])
g = sns.relplot(x='value', y='DIC residuals', data=df_work, col='variable', col_wrap=4, facet_kws={'sharex':False})
_ = g.set_titles(template='{col_name}', row_template='{col_name}')


df_test_ta1 = pd.read_csv('results/bottle_data_test_ta_all_vars.csv', index_col=0)

df_work = pd.melt(df_test_ta1, value_vars=df_test_ta1.columns.tolist()[:8], id_vars=['TA residuals'])
g = sns.relplot(x='value', y='TA residuals', data=df_work, col='variable', col_wrap=4, facet_kws={'sharex':False})
_ = g.set_titles(template='{col_name}', row_template='{col_name}')


df_test_ta2 = pd.read_csv('results/bottle_data_test_ta_noO2.csv', index_col=0)

df_work = pd.melt(df_test_ta2, value_vars=df_test_ta2.columns.tolist()[:7], id_vars=['TA residuals'])
g = sns.relplot(x='value', y='TA residuals', data=df_work, col='variable', col_wrap=4, facet_kws={'sharex':False})
_ = g.set_titles(template='{col_name}', row_template='{col_name}')


df_test_ta3 = pd.read_csv('results/bottle_data_test_ta_noO2.csv', index_col=0)

df_work = pd.melt(df_test_ta3, value_vars=df_test_ta3.columns.tolist()[:4], id_vars=['TA residuals'])
g = sns.relplot(x='value', y='TA residuals', data=df_work, col='variable', col_wrap=4, facet_kws={'sharex':False})
_ = g.set_titles(template='{col_name}', row_template='{col_name}')

Look at residuals of predictions on the test set¶

DIC predictions¶

Model including all available features¶

Model excluding $O_2$¶

Model excluding $O_2$ and satellite data¶

TA predictions¶

Model including all available features¶

Model excluding $O_2$¶

Model excluding $O_2$ and satellite data¶