Ts Notebook

Análisis Tipo de Cambio
El objetivo de este análisis es identificar la influencia de las principales variables economicas en el tipo de cambio USD - DOP.
Variables a analizar con respecto al tipo de cambio USD - DOP:
1. Nivel de Reservas en Millones de USD

2. Tasa Politica Monetaria BCRD
3. Llegada de pasajeros turistas en Miles de Personas
4. Exportaciones en Millones de USD
5. Remesas en Millones de USD
In [1]: import pandas as pd

import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.metrics import mean_absolute_error
from statsmodels.tsa.seasonal import seasonal_decompose
import warnings
warnings.filterwarnings("ignore")
In [2]: df = pd.read_excel(r'M:\MIDDLE-OFFICE\COMUNES\0. GAF\Analisis Tipo de Cambio\DataConsolidada.xlsx', sheet_name='Data')
In [3]: df.head()
Out[3]: Fecha ExportUSDMM ReservUSDMM TPM Turistas Remesas TC
0 2012-01-31 478.892452 3046.8 0.0675 480.044 300.039840 38.858355
1 2012-02-29 503.834791 2980.1 0.0675 474.935 366.658553 38.942515
2 2012-03-31 659.810339 3031.0 0.0675 511.653 409.431621 38.994382
3 2012-04-30 546.674173 3125.5 0.0675 441.475 327.994910 39.017556
4 2012-05-31 661.985431 3012.3 0.0675 352.566 341.065591 39.022152
In [4]: df.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 127 entries, 0 to 126
Data columns (total 7 columns):
# Column Non-Null Count Dtype

--- ------ -------------- -----
0 Fecha 127 non-null datetime64[ns]
1 ExportUSDMM 127 non-null float64
2 ReservUSDMM 127 non-null float64
3 TPM 127 non-null float64
4 Turistas 127 non-null float64
5 Remesas 127 non-null float64
6 TC 127 non-null float64
dtypes: datetime64[ns](1), float64(6)
memory usage: 7.1 KB
In [5]: pd.options.display.float_format = '{:,.2f}'.format

df.describe()
Out[5]: ExportUSDMM ReservUSDMM TPM Turistas Remesas TC
count 127.00 127.00 127.00 127.00 127.00 127.00
mean 770.90 6,802.11 0.05 494.09 526.65 48.10
std 141.38 3,238.95 0.01 155.97 179.84 5.81
min 478.89 2,980.10 0.03 1.38 262.32 38.86
25% 670.21 4,462.30 0.05 422.56 391.58 43.57
50% 756.43 6,248.75 0.05 507.43 474.03 47.28
75% 846.74 7,954.31 0.06 606.86 602.27 52.80
max 1,214.42 14,849.76 0.07 800.94 994.89 58.36
Coeficientes de Variación
Determina la variable con mayor variabilidad
In [6]: var_expor = df.describe()['ExportUSDMM']['std']/df.describe()['ExportUSDMM']['mean']

var_reserv = df.describe()['ReservUSDMM']['std']/df.describe()['ReservUSDMM']['mean']
var_tpm = df.describe()['TPM']['std']/df.describe()['TPM']['mean']
var_tur = df.describe()['Turistas']['std']/df.describe()['Turistas']['mean']
var_rem = df.describe()['Remesas']['std']/df.describe()['ExportUSDMM']['mean']
var = [var_expor,var_reserv,var_tpm,var_tur,var_rem]
var_df = pd.DataFrame(var, index=['Exportaciones','Reservas','TPM','Turistas','Remesas'], columns=['Coeficiente de Variación'])
var_df
Out[6]: Coeficiente de Variación
Exportaciones 0.18
Reservas 0.48
TPM 0.21
Turistas 0.32
Remesas 0.23
In [7]: var_df.sort_values(by='Coeficiente de Variación').plot(kind='barh')
Out[7]: <AxesSubplot:>
Coeficientes de Variación sin periodo Covid
Determina la variable con mayor variabilidad excluyendo el 2020 pandemia.
In [8]: df_nocovid = df[df['Fecha'].dt.year != 2020]
In [9]: var_expor_cov = df_nocovid.describe()['ExportUSDMM']['std']/df.describe()['ExportUSDMM']['mean']

var_reserv_cov = df_nocovid.describe()['ReservUSDMM']['std']/df.describe()['ReservUSDMM']['mean']
var_tpm_cov = df_nocovid.describe()['TPM']['std']/df.describe()['TPM']['mean']
var_tur_cov = df_nocovid.describe()['Turistas']['std']/df.describe()['Turistas']['mean']
var_rem_cov = df_nocovid.describe()['Remesas']['std']/df.describe()['ExportUSDMM']['mean']
var_cov = [var_expor_cov,var_reserv_cov,var_tpm_cov,var_tur_cov,var_rem_cov]
var_df_cov = pd.DataFrame(var_cov, index=['Exportaciones','Reservas','TPM','Turistas','Remesas'], columns=['Coeficiente de Variación'])
var_df_cov
Out[9]: Coeficiente de Variación
Exportaciones 0.19
Reservas 0.48
TPM 0.19
Turistas 0.24
Remesas 0.23
In [10]: var_df_cov.sort_values(by='Coeficiente de Variación').plot(kind='barh')
Graficos de tendencia individual

Se utiliza para analizar el comportamiento de las variables
In [11]: df.set_index('Fecha',inplace=True)
In [12]: #Funcion para automatizar la evaluacion de promedio moviles y valores atipicos

def GraficarPromedioMovil(df,ventana,intervalos=False,escala=2,anomalias=False):
promedio_movil = df.rolling(window=ventana).mean()
plt.figure(figsize=(15,5))
plt.title("Promedio movil\n ventana = {}\n {}".format(ventana,df.columns[0]))
plt.plot(promedio_movil,"g",label="Tendencia de Promedio Movil")
if intervalos:
error_absoluto_medio = mean_absolute_error(df[ventana:],df.rolling(window=ventana).mean()[ventana:])
desviacion_error = np.std(df[ventana:] - df.rolling(window=ventana).mean()[ventana:])
banda_inferior = df.rolling(window=ventana).mean()[ventana:] - (error_absoluto_medio + escala*desviacion_error)
banda_superior = df.rolling(window=ventana).mean()[ventana:] + (error_absoluto_medio + escala*desviacion_error)
plt.plot(banda_superior,"r--", label="Banda Superior /Banda Inferior")
plt.plot(banda_inferior,"r--")
if anomalias:
anomalias = pd.DataFrame(index=df.index, columns=df.columns)
anomalias[df[ventana:]<banda_inferior] = df[ventana:][df[ventana:]<banda_inferior]
anomalias[df[ventana:]>banda_superior] = df[ventana:][df[ventana:]>banda_superior]
plt.plot(anomalias,"ro",markersize=10)
plt.plot(df[ventana:], label="Valores reales")

plt.legend(loc="upper left")
plt.grid(True)
In [13]: exportaciones = pd.DataFrame(df['ExportUSDMM'])

reservas = pd.DataFrame(df['ReservUSDMM'])
tpm = pd.DataFrame(df['TPM'])
turistas = pd.DataFrame(df['Turistas'])
remesas = pd.DataFrame(df['Remesas'])
tc = pd.DataFrame(df['TC'])
Remesas
In [14]: GraficarPromedioMovil(remesas,30,intervalos=True,anomalias=True)
In [15]: decomposition = seasonal_decompose(remesas,period = 30)

tendencia = decomposition.trend
temporalidad = decomposition.seasonal
residuos = decomposition.resid
plt.subplot(411)
plt.plot(remesas, label='Original')
plt.legend(loc='best')
plt.subplot(412)
plt.plot(tendencia, label='Tendencia')
plt.subplot(413)
plt.plot(temporalidad,label='Temporalidad')
plt.subplot(414)
plt.plot(residuos, label='Residuos')
plt.suptitle('Remesas',fontsize=15)
plt.tight_layout()
Exportaciones
In [16]: GraficarPromedioMovil(exportaciones,30,intervalos=True,anomalias=True)
In [17]: decomposition = seasonal_decompose(exportaciones,period = 30)

plt.subplot(411)
plt.plot(exportaciones, label='Original')
plt.subplot(412)
plt.subplot(413)
plt.subplot(414)
plt.suptitle('Exportaciones',fontsize=15)
plt.tight_layout()
Reservas
In [18]: GraficarPromedioMovil(reservas,30,intervalos=True,anomalias=True)
In [19]: decomposition = seasonal_decompose(reservas,period = 30)

plt.subplot(411)
plt.plot(reservas, label='Original')
plt.subplot(412)
plt.subplot(413)
plt.subplot(414)
plt.suptitle('Reservas',fontsize=15)
plt.tight_layout()
Tasa Politica Monetaria

In [20]: GraficarPromedioMovil(tpm,30,intervalos=True,anomalias=True)
In [21]: decomposition = seasonal_decompose(tpm,period = 30)

plt.subplot(411)
plt.plot(tpm, label='Original')
plt.subplot(412)
plt.subplot(413)
plt.subplot(414)
plt.suptitle('Tasa Politica Monetaria',fontsize=15)
plt.tight_layout()
Llegada de Turistas
In [22]: GraficarPromedioMovil(turistas,30,intervalos=True,anomalias=True)
In [23]: decomposition = seasonal_decompose(turistas,period = 30)

plt.subplot(411)
plt.plot(turistas, label='Original')
plt.subplot(412)
plt.subplot(413)
plt.subplot(414)
plt.suptitle('Llegada de Turistas',fontsize=15)
plt.tight_layout()
Tipo de Cambio
In [24]: GraficarPromedioMovil(tc,30,intervalos=True,anomalias=True)
In [25]: decomposition = seasonal_decompose(tc,period = 30)

plt.subplot(411)
plt.plot(tc, label='Original')
plt.subplot(412)
plt.subplot(413)
plt.subplot(414)
plt.suptitle('Tipo de Cambio USD-DOP',fontsize=15)
plt.tight_layout()
Análisis tendencia de variables escala normalizada

Se utiliza para poder comparar las tendencias de todas las variables en una misma escala.
In [26]: from sklearn.preprocessing import MinMaxScaler
In [27]: scaler = MinMaxScaler()

exportaciones['normalizado'] = scaler.fit_transform(exportaciones)

reservas['normalizado'] = scaler.fit_transform(reservas)

tpm['normalizado'] = scaler.fit_transform(tpm)

turistas['normalizado'] = scaler.fit_transform(turistas)

remesas['normalizado'] = scaler.fit_transform(remesas)

tc['normalizado'] = scaler.fit_transform(tc)
In [33]: plt.figure(figsize=(12,6))
plt.plot(reservas['normalizado'], label = 'Reservas')
plt.plot(exportaciones['normalizado'], label = 'Exportaciones')
plt.plot(tpm['normalizado'], label = 'Tasa Politica Monetaria')
plt.plot(turistas['normalizado'], label = 'Turistas')
plt.plot(remesas['normalizado'], label = 'Remesas')
plt.plot(tc['normalizado'], label = 'Tasa de Cambio USD-DOP', color='b')
plt.show()
In [34]: plt.plot(remesas['normalizado'], label = 'Remesas')

plt.suptitle('Remesas')
plt.show()
In [35]: plt.plot(reservas['normalizado'], label = 'Reservas')

plt.suptitle('Reservas')
plt.show()
In [36]: plt.plot(exportaciones['normalizado'], label = 'Exportaciones')

plt.suptitle('Exportaciones')
plt.show()
In [37]: plt.plot(tpm['normalizado'], label = 'Tasa Politica Monetaria')
plt.suptitle('Tasa Politica Monetaria')
plt.show()
In [38]: .plt.plot(turistas['normalizado'], label = 'Turistas')

plt.suptitle('Llegada de Turistas')
plt.show()
Análisis de relación de variables

Se utilizara poder analizar cuales variables están correlacionadas y que tanto.
In [39]: import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
import seaborn as sns
import numpy as np

class SeabornFig2Grid():

def __init__(self, seaborngrid, fig, subplot_spec):
self.fig = fig
self.sg = seaborngrid
self.subplot = subplot_spec
if isinstance(self.sg, sns.axisgrid.FacetGrid) or \
isinstance(self.sg, sns.axisgrid.PairGrid):
self._movegrid()
elif isinstance(self.sg, sns.axisgrid.JointGrid):
self._movejointgrid()
self._finalize()

def _movegrid(self):
""" Move PairGrid or Facetgrid """
self._resize()
n = self.sg.axes.shape[0]
m = self.sg.axes.shape[1]
self.subgrid = gridspec.GridSpecFromSubplotSpec(n,m, subplot_spec=self.subplot)
for i in range(n):
for j in range(m):
self._moveaxes(self.sg.axes[i,j], self.subgrid[i,j])

def _movejointgrid(self):
""" Move Jointgrid """
h= self.sg.ax_joint.get_position().height
h2= self.sg.ax_marg_x.get_position().height
r = int(np.round(h/h2))
self._resize()
self.subgrid = gridspec.GridSpecFromSubplotSpec(r+1,r+1, subplot_spec=self.subplot)

self._moveaxes(self.sg.ax_joint, self.subgrid[1:, :-1])
self._moveaxes(self.sg.ax_marg_x, self.subgrid[0, :-1])
self._moveaxes(self.sg.ax_marg_y, self.subgrid[1:, -1])

def _moveaxes(self, ax, gs):
#https://stackoverflow.com/a/46906599/4124317
ax.remove()
ax.figure=self.fig
self.fig.axes.append(ax)
self.fig.add_axes(ax)
ax._subplotspec = gs
ax.set_position(gs.get_position(self.fig))
ax.set_subplotspec(gs)

def _finalize(self):
plt.close(self.sg.fig)
self.fig.canvas.mpl_connect("resize_event", self._resize)
self.fig.canvas.draw()

def _resize(self, evt=None):
self.sg.fig.set_size_inches(self.fig.get_size_inches())
In [40]: sns.jointplot(data=df, x="Remesas", y="TC")

plt.suptitle("Relación Remesas vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['Remesas'],df['TC'])[0][1],2)),y = 1)
plt.show()
In [41]: sns.jointplot(data=df, x="Remesas", y="TC", hue=df.index.year,palette="Paired")

plt.suptitle("Relación Remesas vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['Remesas'],df['TC'])[0][1],2)),y = 1)
plt.show()
In [42]: g0 = sns.jointplot(data=df, x="Remesas", y="TC")
g1 = sns.jointplot(data=df, x="Remesas", y="TC", hue=df.index.year)

fig = plt.figure(figsize=(7,5))
gs = gridspec.GridSpec(1, 2)
plt.suptitle("Relación Remesas vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['Remesas'],df['TC'])[0][1],2)))

mg0 = SeabornFig2Grid(g0, fig, gs[0])
In [43]: sns.jointplot(data=df, x="ExportUSDMM", y="TC")

plt.suptitle("Relación Exportaciones vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['ExportUSDMM'],df['TC'])[0][1],2)), y=1)
plt.show()
In [44]: sns.jointplot(data=df, x="ExportUSDMM", y="TC", hue=df.index.year,palette="Paired")

plt.suptitle("Relación Exportaciones vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['ExportUSDMM'],df['TC'])[0][1],2)),y = 1)
plt.show()
In [45]: g0 = sns.jointplot(data=df, x="ExportUSDMM", y="TC")
g1 = sns.jointplot(data=df, x="ExportUSDMM", y="TC", hue=df.index.year,palette="Paired")

plt.suptitle("Relación Exportaciones vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['ExportUSDMM'],df['TC'])[0][1],2)))

In [46]: sns.jointplot(data=df, x="ReservUSDMM", y="TC", hue=df.index.year,palette="Paired")

plt.suptitle("Relación Reservas vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['ReservUSDMM'],df['TC'])[0][1],2)),y = 1)
plt.show()
In [47]: sns.jointplot(data=df, x="ReservUSDMM", y="TC")

plt.suptitle("Relación Reservas vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['ReservUSDMM'],df['TC'])[0][1],2)),y = 1)
plt.show()
In [48]: g0 = sns.jointplot(data=df, x="ReservUSDMM", y="TC")
g1 = sns.jointplot(data=df, x="ReservUSDMM", y="TC", hue=df.index.year,palette="Paired")

plt.suptitle("Relación Reservas vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['ReservUSDMM'],df['TC'])[0][1],2)))

In [49]: sns.jointplot(data=df, x="TPM", y="TC")

plt.suptitle("Relación TPM vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['TPM'],df['TC'])[0][1],2)),y = 1)
plt.show()
In [50]: sns.jointplot(data=df, x="TPM", y="TC", hue=df.index.year,palette="Paired")

plt.suptitle("Relación TPM vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['TPM'],df['TC'])[0][1],2)),y = 1)
plt.show()
In [51]: g0 = sns.jointplot(data=df, x="TPM", y="TC")
g1 = sns.jointplot(data=df, x="TPM", y="TC", hue=df.index.year,palette="Paired")

plt.suptitle("Relación TPM vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['TPM'],df['TC'])[0][1],2)))

In [52]: sns.jointplot(data=df, x="Turistas", y="TC", hue=df.index.year,palette="Paired")

plt.suptitle("Relación Turistas vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['Turistas'],df['TC'])[0][1],2)),y = 1)
plt.show()
In [53]: sns.jointplot(data=df, x="Turistas", y="TC")

plt.suptitle("Relación Turistas vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['Turistas'],df['TC'])[0][1],2)),y = 1)
plt.show()
In [54]: g0 = sns.jointplot(data=df, x="Turistas", y="TC")
g1 = sns.jointplot(data=df, x="Turistas", y="TC", hue=df.index.year,palette="Paired")

plt.suptitle("Relación Llegada de Turistas vs Tasa de Cambios USD-DOP Coeficiente de Correlación: {}".format(round(np.corrcoef(df['Turistas'],df['TC'])[0][1],2)))

mg0 = SeabornFig2Grid(g0, fig, gs[0])#Sin Covid
In [55]: def hexbin(x, y, color, **kwargs):

cmap = sns.light_palette(color, as_cmap=True)
plt.hexbin(x, y, gridsize=15, cmap=cmap, extent=[min(x), max(x), min(y), max(y)], **kwargs)

g = sns.PairGrid(df)
g.map_upper(hexbin)
g.map_diag(sns.histplot)
g.map_lower(sns.kdeplot)
g.add_legend()
plt.show()
In [56]: df_orig = df.copy()

df_orig['anio'] = df.index.year
In [57]: g = sns.PairGrid(df_orig, hue = 'anio',palette="Paired")
g.map_upper(sns.scatterplot)
g.map_diag(sns.kdeplot)
g.map_lower(sns.kdeplot)
g.add_legend()
plt.show()
Mapa de calor
In [58]: plt.figure(figsize=(10,6))
sns.heatmap(df.corr(), annot=True,cmap='coolwarm')
In [59]: df.corr()['TC'].sort_values(ascending=False)[1:].plot(kind='bar')
In [60]: df.corr()['TC'].sort_values(ascending=False)[1:]
Out[60]: Remesas 0.91
ReservUSDMM 0.90
ExportUSDMM 0.77
Turistas -0.08
TPM -0.70
Name: TC, dtype: float64
Análisis de Variabilidad a traves del tiempo

Con esto podemos comparar la variabilidad en los diferentes años
In [61]: sns.violinplot(y='Remesas',x='anio',data=df_orig)
plt.suptitle('Remesas')
plt.show()
In [62]: sns.violinplot(y='ExportUSDMM',x='anio',data=df_orig)
plt.suptitle('Exportaciones')
plt.show()
In [63]: sns.violinplot(y='ReservUSDMM',x='anio',data=df_orig)
plt.suptitle('Reservas')
plt.show()
In [64]: sns.violinplot(y='TPM',x='anio',data=df_orig)
plt.suptitle('TPM')
plt.show()
In [65]: sns.violinplot(y='Turistas',x='anio',data=df_orig)
plt.suptitle('Llegada de Turistas')
plt.show()
In [66]: sns.violinplot(y='TC',x='anio',data=df_orig)
plt.suptitle('Tasa de Cambios USD-DOP')
plt.show()

df_orig.groupby(['anio']).agg({'ExportUSDMM':['min','mean','std','max']}
).transpose().style.background_gradient(cmap='RdYlGn_r',axis='columns').format("{:,.2f}")
Out[67]: anio 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
min 478.89 542.83 602.19 563.79 537.73 581.19 677.30 745.73 541.26 740.37 812.01
mean 599.64 661.30 707.49 693.54 728.50 735.93 785.85 839.93 820.40 970.38 1,055.56
ExportUSDMM
std 76.04 68.69 58.36 63.80 74.43 77.49 57.31 55.69 114.88 96.38 122.35
max 757.49 782.42 786.56 800.23 796.22 843.28 873.43 921.50 924.04 1,114.46 1,214.42
In [68]: df_orig.groupby(['anio']).agg({'Remesas':['min','mean','std','max']}
Out[68]: anio 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
min 263.52 262.32 288.50 338.76 355.07 440.93 455.88 537.03 395.00 760.96 748.81
mean 337.11 355.19 380.94 413.40 438.40 492.65 541.17 590.59 684.94 866.87 809.95
Remesas
std 38.41 54.84 47.78 47.60 43.78 41.41 45.34 46.16 140.25 71.68 48.57
max 409.43 436.61 459.76 486.89 504.03 563.90 603.93 665.49 872.31 994.89 888.13
In [69]: df_orig.groupby(['anio']).agg({'ReservUSDMM':['min','mean','std','max']}
Out[69]: anio 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
min 2,980.10 3,017.80 3,449.50 4,369.80 4,938.60 5,821.83 6,597.96 6,980.51 6,689.18 11,963.16 12,373.98
mean 3,092.84 3,610.73 4,258.23 4,770.82 5,314.52 6,335.23 7,215.96 7,635.21 9,012.52 12,484.17 14,157.91
ReservUSDMM
std 94.75 420.73 500.26 212.92 339.35 260.71 433.00 630.75 1,469.00 366.65 817.88
max 3,245.15 4,386.50 5,162.00 5,195.05 6,046.70 6,780.40 8,050.15 8,781.40 10,751.62 13,060.29 14,849.76
In [70]: df_orig.groupby(['anio']).agg({'TPM':['min','mean','std','max']}
Out[70]: anio 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
min 0.05 0.04 0.06 0.05 0.05 0.05 0.05 0.04 0.03 0.03 0.04
mean 0.06 0.05 0.06 0.05 0.05 0.05 0.05 0.05 0.04 0.03 0.06
TPM
std 0.01 0.01 0.00 0.01 0.00 0.00 0.00 0.00 0.01 0.00 0.01
max 0.07 0.06 0.06 0.06 0.06 0.06 0.06 0.06 0.04 0.04 0.07
In [71]: df_orig.groupby(['anio']).agg({'Turistas':['min','mean','std','max']}
Out[71]: anio 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
min 298.53 287.63 324.69 373.73 404.99 360.46 422.15 385.02 1.38 224.00 607.40
mean 420.59 430.31 470.73 512.58 546.58 569.32 601.69 593.90 225.62 465.84 671.83
Turistas
std 74.99 79.56 75.50 76.66 82.85 96.62 95.09 103.81 217.78 159.98 67.48
max 511.65 537.04 581.48 633.71 679.39 702.86 731.82 714.09 636.88 788.41 800.94
In [72]: df_orig.groupby(['anio']).agg({'TC':['min','mean','std','max']}
Out[72]: anio 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
min 38.86 40.58 42.86 44.46 45.53 46.69 48.33 50.29 53.04 56.27 54.47
mean 39.24 41.69 43.44 44.93 45.98 47.44 49.43 51.19 56.41 56.96 55.42
TC
std 0.41 0.79 0.40 0.30 0.34 0.38 0.53 1.00 2.30 0.53 1.11
max 40.17 42.61 44.14 45.44 46.59 48.11 50.15 52.84 58.36 58.07 57.52
Análisis de regresión multiple con Tipo de Cambio

In [73]: X = df.drop('TC', axis=1)
y = df['TC']
In [75]: x_scale = scaler.fit_transform(X)
In [76]: df_scale = pd.DataFrame(x_scale, columns=X.columns)
In [77]: from sklearn.model_selection import train_test_split
In [78]: X_train, X_test, y_train, y_test = train_test_split(df_scale, y, test_size=0.15, random_state=42)
In [79]: from sklearn.linear_model import LinearRegression
In [80]: lr = LinearRegression()
In [81]: lr.fit(X_train,y_train)
Out[81]: LinearRegression()
In [82]: lr.intercept_
Out[82]: 45.359216989988944

df_coef = pd.DataFrame(lr.coef_, index=X.columns, columns=['Coeficientes'])
df_coef
Out[83]: Coeficientes
ExportUSDMM 0.204056
ReservUSDMM 9.004129
TPM -3.137765
Turistas -4.022236
Remesas 10.493490
In [84]: predicciones = lr.predict(X_test)
In [85]: plt.scatter(y_test,predicciones)
Out[85]: <matplotlib.collections.PathCollection at 0x20451b5c3d0>
In [86]: residuos = y_test - predicciones

sns.histplot(residuos)
Out[86]: <AxesSubplot:xlabel='TC', ylabel='Count'>

In [87]: comparacion = pd.concat([y.reset_index(drop=True),pd.Series(lr.predict(df_scale))], axis=1)
comparacion.columns = ['TC','Estimaciones']
comparacion.index = df.index
In [88]: comparacion.plot()
plt.show()
In [89]: from sklearn.metrics import mean_absolute_error, mean_squared_error
In [90]: print('Error Absoluto Medio: ', mean_absolute_error(y_test,predicciones))

print('Error Cuadratico Medio: ', mean_absolute_error(y_test,predicciones))
print('Raiz Error Cuadratico Medio: ', np.sqrt(mean_absolute_error(y_test,predicciones)))
Error Absoluto Medio: 1.6393960140099282
Error Cuadratico Medio: 1.6393960140099282
Raiz Error Cuadratico Medio: 1.28038900886017
In [91]: import statsmodels.api as sm
In [92]: X = sm.add_constant(df_scale)
modelo = sm.OLS(y.reset_index(drop=True),X)
result = modelo.fit()
print(result.summary())
OLS Regression Results
==============================================================================
Dep. Variable: TC R-squared: 0.899
Model: OLS Adj. R-squared: 0.894
Method: Least Squares F-statistic: 214.5
Date: Tue, 06 Sep 2022 Prob (F-statistic): 2.24e-58
Time: 12:00:54 Log-Likelihood: -257.84
No. Observations: 127 AIC: 527.7
Df Residuals: 121 BIC: 544.7
Df Model: 5
Covariance Type: nonrobust
===============================================================================
coef std err t P>|t| [0.025 0.975]
-------------------------------------------------------------------------------
const 45.2150 0.768 58.911 0.000 43.695 46.734
ExportUSDMM -0.4241 1.687 -0.251 0.802 -3.763 2.915
ReservUSDMM 10.2088 1.613 6.330 0.000 7.016 13.402
TPM -3.0153 1.071 -2.815 0.006 -5.136 -0.894
Turistas -3.7357 1.057 -3.533 0.001 -5.829 -1.642
Remesas 9.7854 1.982 4.937 0.000 5.862 13.709
==============================================================================
Omnibus: 5.312 Durbin-Watson: 0.346
Prob(Omnibus): 0.070 Jarque-Bera (JB): 4.010
Skew: -0.310 Prob(JB): 0.135
Kurtosis: 2.390 Cond. No. 20.5
==============================================================================
Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
Análisis de regresión multiple con Tipo de Cambio sin pandemia

In [93]: X = df[df.index.year !=2020].drop('TC', axis=1)
y = df[df.index.year !=2020]['TC']
In [95]: x_scale = scaler.fit_transform(X)
In [96]: df_scale = pd.DataFrame(x_scale, columns=X.columns)
In [97]: from sklearn.model_selection import train_test_split
In [98]: X_train, X_test, y_train, y_test = train_test_split(x_scale, y, test_size=0.15, random_state=42)
In [99]: from sklearn.linear_model import LinearRegression
In [100]: lr = LinearRegression()
In [101]: lr.fit(X_train,y_train)
Out[101]: LinearRegression()
In [102]: lr.intercept_
Out[102]: 42.94967306067477

df_coef = pd.DataFrame(lr.coef_, index=X.columns, columns=['Coeficientes'])
df_coef
Out[103]: Coeficientes
ExportUSDMM 1.704006
ReservUSDMM 12.178299
TPM -3.544393
Turistas 1.030427
Remesas 3.798929
In [104]: predicciones = lr.predict(X_test)
In [105]: plt.scatter(y_test,predicciones)
Out[105]: <matplotlib.collections.PathCollection at 0x20454d6f820>
In [106]: residuos = y_test - predicciones

sns.histplot(residuos)
Out[106]: <AxesSubplot:xlabel='TC', ylabel='Count'>
In [107]: from sklearn.metrics import mean_absolute_error, mean_squared_error
In [108]: print('Error Absoluto Medio: ', mean_absolute_error(y_test,predicciones))

print('Error Cuadratico Medio: ', mean_absolute_error(y_test,predicciones))
print('Raiz Error Cuadratico Medio: ', np.sqrt(mean_absolute_error(y_test,predicciones)))
Error Absoluto Medio: 1.428469176476815
Error Cuadratico Medio: 1.428469176476815
Raiz Error Cuadratico Medio: 1.195185833448847
In [109]: X = sm.add_constant(df_scale)
modelo = sm.OLS(y.reset_index(drop=True),X)
result = modelo.fit()
print(result.summary())
OLS Regression Results
==============================================================================
Dep. Variable: TC R-squared: 0.912
Model: OLS Adj. R-squared: 0.908
Method: Least Squares F-statistic: 226.4
Date: Tue, 06 Sep 2022 Prob (F-statistic): 7.41e-56
Time: 12:00:54 Log-Likelihood: -216.00
No. Observations: 115 AIC: 444.0
Df Residuals: 109 BIC: 460.5
Df Model: 5
Covariance Type: nonrobust
===============================================================================
coef std err t P>|t| [0.025 0.975]
-------------------------------------------------------------------------------
const 42.8353 0.741 57.773 0.000 41.366 44.305
ExportUSDMM 1.4992 1.555 0.964 0.337 -1.583 4.582
ReservUSDMM 12.5113 1.832 6.828 0.000 8.879 16.143
TPM -3.5846 0.970 -3.697 0.000 -5.506 -1.663
Turistas 1.3659 0.901 1.516 0.132 -0.420 3.152
Remesas 3.5417 2.318 1.528 0.130 -1.053 8.137
==============================================================================
Omnibus: 2.025 Durbin-Watson: 0.387
Prob(Omnibus): 0.363 Jarque-Bera (JB): 1.879
Skew: -0.216 Prob(JB): 0.391
Kurtosis: 2.547 Cond. No. 26.1
==============================================================================
Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
Conclusión
Historicamente las variables seleccionadas tienen una correlación positiva con la variable objetivo, a excepción de la TPM
Las unicas variables que influyen negativamente, es decir, que aportan a la disminucion de la tasa de cambio, es la TPM y la llegada de turistas
Aunque no podemos evidenciar que las demas variables contribuyen positivamente (causalidad/correlacion) a la tasa de cambio, podemos notar que su incremento no crea una disminución, quizás contribuya a desacelerar el aumento,
pero los niveles observados no han contribuido a la disminución como tal

Ts Notebook

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ts Notebook

Uploaded by

Copyright:

Available Formats

Análisis Tipo de Cambio

Variables a analizar con respecto al tipo de cambio USD - DOP:

1. Nivel de Reservas en Millones de USD

In [1]: import pandas as pd

In [2]: df = pd.read_excel(r'M:\MIDDLE-OFFICE\COMUNES\0. GAF\Analisis Tipo de Cambio\DataConsolidada.xlsx', sheet_name='Data')

Out[3]: Fecha ExportUSDMM ReservUSDMM TPM Turistas Remesas TC

0 2012-01-31 478.892452 3046.8 0.0675 480.044 300.039840 38.858355

1 2012-02-29 503.834791 2980.1 0.0675 474.935 366.658553 38.942515

2 2012-03-31 659.810339 3031.0 0.0675 511.653 409.431621 38.994382

3 2012-04-30 546.674173 3125.5 0.0675 441.475 327.994910 39.017556

4 2012-05-31 661.985431 3012.3 0.0675 352.566 341.065591 39.022152

RangeIndex: 127 entries, 0 to 126

Data columns (total 7 columns):

# Column Non-Null Count Dtype

memory usage: 7.1 KB

In [5]: pd.options.display.float_format = '{:,.2f}'.format

Out[5]: ExportUSDMM ReservUSDMM TPM Turistas Remesas TC

count 127.00 127.00 127.00 127.00 127.00 127.00

mean 770.90 6,802.11 0.05 494.09 526.65 48.10

std 141.38 3,238.95 0.01 155.97 179.84 5.81

min 478.89 2,980.10 0.03 1.38 262.32 38.86

25% 670.21 4,462.30 0.05 422.56 391.58 43.57

50% 756.43 6,248.75 0.05 507.43 474.03 47.28

75% 846.74 7,954.31 0.06 606.86 602.27 52.80

max 1,214.42 14,849.76 0.07 800.94 994.89 58.36

In [6]: var_expor = df.describe()['ExportUSDMM']['std']/df.describe()['ExportUSDMM']['mean']

Out[6]: Coeficiente de Variación

In [7]: var_df.sort_values(by='Coeficiente de Variación').plot(kind='barh')

In [8]: df_nocovid = df[df['Fecha'].dt.year != 2020]

In [9]: var_expor_cov = df_nocovid.describe()['ExportUSDMM']['std']/df.describe()['ExportUSDMM']['mean']

Out[9]: Coeficiente de Variación

In [10]: var_df_cov.sort_values(by='Coeficiente de Variación').plot(kind='barh')

Graficos de tendencia individual

In [12]: #Funcion para automatizar la evaluacion de promedio moviles y valores atipicos

plt.plot(df[ventana:], label="Valores reales")

In [13]: exportaciones = pd.DataFrame(df['ExportUSDMM'])

Tasa Politica Monetaria

Análisis tendencia de variables escala normalizada

In [26]: from sklearn.preprocessing import MinMaxScaler

In [27]: scaler = MinMaxScaler()

In [28]: scaler = MinMaxScaler()

In [29]: scaler = MinMaxScaler()

In [30]: scaler = MinMaxScaler()

In [31]: scaler = MinMaxScaler()

In [34]: plt.plot(remesas['normalizado'], label = 'Remesas')

In [35]: plt.plot(reservas['normalizado'], label = 'Reservas')

In [36]: plt.plot(exportaciones['normalizado'], label = 'Exportaciones')

In [38]: .plt.plot(turistas['normalizado'], label = 'Turistas')

Análisis de relación de variables

In [40]: sns.jointplot(data=df, x="Remesas", y="TC")

In [41]: sns.jointplot(data=df, x="Remesas", y="TC", hue=df.index.year,palette="Paired")

In [43]: sns.jointplot(data=df, x="ExportUSDMM", y="TC")

In [44]: sns.jointplot(data=df, x="ExportUSDMM", y="TC", hue=df.index.year,palette="Paired")

In [46]: sns.jointplot(data=df, x="ReservUSDMM", y="TC", hue=df.index.year,palette="Paired")

In [47]: sns.jointplot(data=df, x="ReservUSDMM", y="TC")

In [49]: sns.jointplot(data=df, x="TPM", y="TC")

In [50]: sns.jointplot(data=df, x="TPM", y="TC", hue=df.index.year,palette="Paired")

In [52]: sns.jointplot(data=df, x="Turistas", y="TC", hue=df.index.year,palette="Paired")

In [53]: sns.jointplot(data=df, x="Turistas", y="TC")

In [55]: def hexbin(x, y, color, **kwargs):

In [56]: df_orig = df.copy()

Out[60]: Remesas 0.91

Name: TC, dtype: float64

Análisis de Variabilidad a traves del tiempo