程序代做 # Impor􏰁 Librarie􏰂

# Impor􏰁 Librarie􏰂
im􏰃􏰄􏰅􏰁 n􏰆mp􏰇 a􏰂 np # package for 􏰂cien􏰁ific comp􏰆􏰁ing
im􏰃􏰄􏰅􏰁 pandas a􏰂 pd # package for da􏰁a manip􏰆la􏰁ion
im􏰃􏰄􏰅􏰁 os # package for comm􏰆nica􏰁ing 􏰈i􏰁h opera􏰁ing 􏰂􏰇􏰂􏰁em

im􏰃􏰄􏰅􏰁 statsmodels.form􏰆la.api a􏰂 smf # package for 􏰂􏰁a􏰁i􏰂􏰁ical model􏰂 i.e. OLS im􏰃􏰄􏰅􏰁 scip􏰇.stats a􏰂 st
# Read da􏰁a
df = pd.E􏰉celFile(‘Ta􏰇lor r􏰆le f􏰆ndamentals data.􏰉ls’) japan = pd.read_e􏰉cel(df, ‘Japan’)
japan.head(10)
### Da􏰁a Cleaning : U􏰂e ob􏰂er􏰊a􏰁ion􏰂 from 1973M3 􏰁o 2006M6
## Dependen􏰁 Variable : 􏰂_􏰁+1, 􏰂_􏰁
# 􏰂_c􏰆rren􏰁 : log of nominal e􏰉change ra􏰁e􏰂 a􏰁 􏰁
s_c􏰆rrent = np.log(japan.loc[26:len(japan)-2,’EX_JAP_2′]).reset_inde􏰉(drop=T􏰅􏰆e) #Wh􏰇 re􏰂e􏰁 inde􏰉? To combine in􏰁o a 􏰂ingle da􏰁a 􏰂e􏰁 la􏰁 s_c􏰆rrent = s_c􏰆rrent.rename(‘s_c􏰆rrent’) #make 􏰂􏰆re 􏰁hing􏰂 are named 􏰁he 􏰈a􏰇 􏰈e 􏰈an􏰁 􏰁hem.
#I 􏰆􏰂ed and 􏰈ill 􏰆􏰂e “26” abo􏰊e beca􏰆􏰂e i j􏰆􏰂􏰁 􏰈an􏰁 􏰁he da􏰁a 􏰂􏰁ar􏰁ing from 1973,􏰈hich 􏰈hen I looked a􏰁 􏰁he e􏰉cel file I 􏰂a􏰈 􏰂􏰁ar􏰁ed a􏰁 r #Of co􏰆r􏰂e, 􏰇o􏰆 can al􏰈a􏰇􏰂 􏰂􏰆b􏰂e􏰁 􏰁he da􏰁a􏰂e􏰁 􏰁o e􏰉cl􏰆de e􏰊er􏰇􏰁hing from before 1973.Ho􏰈e􏰊er, 􏰈e al􏰂o need da􏰁a 􏰁ha􏰁 came before 1973 􏰁o

## E􏰂􏰁ima􏰁e of O􏰆􏰁p􏰆􏰁 Gap
# Linear Trend
# O􏰆􏰁p􏰆􏰁 Gap in Japan and US
japan[‘t’] = pd.DataFrame(􏰋’t’ : range(1,len(japan)+1)􏰌) #j􏰆􏰂􏰁 􏰆􏰂e 􏰂imple 􏰁ime for 􏰁he linear 􏰁rend
f􏰄􏰅 i i􏰍 range(0,len(japan_const)): # Upda􏰁ing 􏰁he po􏰁en􏰁ial o􏰆􏰁p􏰆􏰁 each period
linear_trend_japan = smf.ols(form􏰆la = ‘IPSA_JAP 􏰎 t’, data=japan[0:27+i]).fit() #gdp regre􏰂􏰂ed again􏰂􏰁 􏰁ime japan_const.loc[i,’japan_potential’] = linear_trend_japan.predict(japan[‘t’][27+i:])[27+i] # 􏰈e need 􏰁he 􏰂econd [27+i] 􏰁o en􏰂􏰆re 􏰁he linear_trend_􏰆s = smf.ols(form􏰆la = ‘IPSA_US 􏰎 t’, data=japan[0:27+i]).fit()
japan_const.loc[i,’􏰆s_potential’] = linear_trend_􏰆s.predict(japan[‘t’][27+i:])[27+i]
japan_const[‘japan_gap’] = (japan_const[‘IPSA_JAP’] – japan_const[‘japan_potential’])/japan_const[‘japan_potential’] japan_const[‘􏰆s_gap’] = (japan_const[‘IPSA_US’] – japan_const[‘􏰆s_potential’])/japan_const[‘􏰆s_potential’]
# O􏰆􏰁p􏰆􏰁 Gap differen􏰁ial be􏰁􏰈een US and Japan
japan_const[‘gap_diff’] = japan_const[‘􏰆s_gap’] – japan_const[‘japan_gap’] japan_const.head(10)
# 􏰂_f􏰆􏰁􏰆re : log of nominal e􏰉change ra􏰁e􏰂 a􏰁 􏰁+1
s_f􏰆t􏰆re = np.log(japan.loc[27:,’EX_JAP_2′]).reset_inde􏰉(drop=T􏰅􏰆e) s_f􏰆t􏰆re = s_f􏰆t􏰆re.rename(‘s_f􏰆t􏰆re’)
# 􏰂_change : change in log of e􏰉change ra􏰁e􏰂 (􏰂_􏰁+1 – 􏰂_􏰁), f􏰆􏰁􏰆re – c􏰆rren􏰁
s_change = s_f􏰆t􏰆re – s_c􏰆rrent s_change = s_change.rename(‘s_change’)
## Independen􏰁 Variable􏰂
# inf_􏰆􏰂, inf_japan : 􏰆􏰂 infla􏰁ion a􏰁 􏰁, japan infla􏰁ion a􏰁 􏰁
inf_japan = japan.loc[26:len(japan)-2,’INFL_JAP’].reset_inde􏰉(drop=T􏰅􏰆e) inf_􏰆s = japan.loc[26:len(japan)-2,’INFL_US’].reset_inde􏰉(drop=T􏰅􏰆e) inf_diff = inf_􏰆s – inf_japan
inf_diff = inf_diff.rename(‘inf_diff’) #infla􏰁ion difference
# o􏰆􏰁p􏰆􏰁_􏰆􏰂, o􏰆􏰁p􏰆􏰁_japan : 􏰆􏰂 o􏰆􏰁p􏰆􏰁 a􏰁 􏰁, japan o􏰆􏰁p􏰆􏰁 a􏰁 􏰁 (GDP)
o􏰆tp􏰆t_japan = japan.loc[26:len(japan)-2,’IPSA_JAP’].reset_inde􏰉(drop=T􏰅􏰆e) o􏰆tp􏰆t_􏰆s = japan.loc[26:len(japan)-2,’IPSA_US’].reset_inde􏰉(drop=T􏰅􏰆e)
# da􏰁e : Time a􏰁 􏰁+1
date = japan.loc[27:,’date’].reset_inde􏰉(drop=T􏰅􏰆e)
# Combine all 􏰊ariable􏰂 in􏰁o a ne􏰈 da􏰁a 􏰂e􏰁
japan_const = pd.concat([date, s_f􏰆t􏰆re, s_c􏰆rrent, s_change, inf_diff, o􏰆tp􏰆t_japan, o􏰆tp􏰆t_􏰆s], a􏰉is=1) japan_const.head(10)

# Q􏰆adra􏰁ic Trend O􏰆􏰁p􏰆􏰁 Gap in Japan and US
japan[‘X’] = japan[“t”]^2 #􏰈i􏰁h 􏰁he q􏰆adra􏰁ic 􏰁rend 􏰈e j􏰆􏰂􏰁 􏰂q􏰆are 􏰁ime
f􏰄􏰅 i i􏰍 range(0,len(japan_const)): # Upda􏰁ing 􏰁he po􏰁en􏰁ial o􏰆􏰁p􏰆􏰁 each period
q􏰆ad_trend_japan = smf.ols(form􏰆la = ‘IPSA_JAP 􏰎 X’, data=japan[0:27+i]).fit() japan_const.loc[i,’japan_potential_q􏰆ad’] = q􏰆ad_trend_japan.predict(japan[‘X’][27+i:])[27+i] q􏰆ad_trend_􏰆s = smf.ols(form􏰆la = ‘IPSA_US 􏰎 X’, data=japan[0:27+i]).fit() japan_const.loc[i,’􏰆s_potential_q􏰆ad’] = q􏰆ad_trend_􏰆s.predict(japan[‘X’][27+i:])[27+i]
japan_const[‘japan_gap_q􏰆ad’] = (japan_const[‘IPSA_JAP’] – japan_const[‘japan_potential_q􏰆ad’])/japan_const[‘japan_potential_q􏰆ad’] japan_const[‘􏰆s_gap_q􏰆ad’] = (japan_const[‘IPSA_US’] – japan_const[‘􏰆s_potential_q􏰆ad’])/japan_const[‘􏰆s_potential_q􏰆ad’]
# O􏰆􏰁p􏰆􏰁 Gap differen􏰁ial be􏰁􏰈een US and Japan
japan_const[‘gap_diff_q􏰆ad’] = japan_const[‘􏰆s_gap_q􏰆ad’] – japan_const[‘japan_gap_q􏰆ad’] # Keep 􏰊ariable􏰂 onl􏰇 􏰆􏰂ed for o􏰆r model
#japan_con􏰂􏰁=japan_con􏰂􏰁[[‘da􏰁e’, ‘􏰂_f􏰆􏰁􏰆re’, ‘􏰂_c􏰆rren􏰁’, ‘􏰂_change’, ‘inf_diff’, ‘gap_diff’, ‘gap_diff_q􏰆ad’]]
japan_const.head(10)
# HP Fil􏰁er Trend O􏰆􏰁p􏰆􏰁 Gap in Japan and US
im􏰃􏰄􏰅􏰁 statsmodels.api a􏰂 sm
c􏰇cle_JAP, trend_JAP = sm.tsa.filters.hpfilter(japan[‘IPSA_JAP’],14400)
c􏰇cle_US, trend_US = sm.tsa.filters.hpfilter(japan[‘IPSA_US’],14400) japan[‘trend_JAP’] = trend_JAP
japan[‘trend_US’] = trend_US
f􏰄􏰅 i i􏰍 range(0,len(japan_const)): # Upda􏰁ing 􏰁he po􏰁en􏰁ial o􏰆􏰁p􏰆􏰁 each period
hp_trend_japan = smf.ols(form􏰆la = ‘IPSA_JAP 􏰎 trend_JAP’, data=japan[0:27+i]).fit() japan_const.loc[i,’japan_potential_hp’] = hp_trend_japan.predict(japan[‘trend_JAP’][27+i:])[27+i] hp_trend_􏰆s = smf.ols(form􏰆la = ‘IPSA_US 􏰎 trend_US’, data=japan[0:27+i]).fit() japan_const.loc[i,’􏰆s_potential_hp’] = hp_trend_􏰆s.predict(japan[‘trend_US’][27+i:])[27+i]
japan_const[‘japan_gap_hp’] = (japan_const[‘IPSA_JAP’] – japan_const[‘japan_potential_hp’])/japan_const[‘japan_potential_hp’] japan_const[‘􏰆s_gap_hp’] = (japan_const[‘IPSA_US’] – japan_const[‘􏰆s_potential_hp’])/japan_const[‘􏰆s_potential_hp’]
# O􏰆􏰁p􏰆􏰁 Gap differen􏰁ial be􏰁􏰈een US and Japan
japan_const[‘gap_diff_hp’] = japan_const[‘􏰆s_gap_hp’] – japan_const[‘japan_gap_hp’] # Keep 􏰊ariable􏰂 onl􏰇 􏰆􏰂ed for o􏰆r model
japan_const=japan_const[[‘date’, ‘s_f􏰆t􏰆re’, ‘s_c􏰆rrent’, ‘s_change’, ‘inf_diff’, ‘gap_diff’, ‘gap_diff_q􏰆ad’, ‘gap_diff_hp’]] japan_const.head(10)

## Make 􏰂􏰆re all col􏰆mn􏰂 ha􏰊e floa􏰁 n􏰆mber􏰂, meaning 􏰁ha􏰁 all n􏰆mber􏰂 con􏰁ain decimal poin􏰁􏰂
japan_const.loc[:,’s_f􏰆t􏰆re’] = japan_const.loc[:,’s_f􏰆t􏰆re’].appl􏰇(float) japan_const.loc[:,’s_c􏰆rrent’] = japan_const.loc[:,’s_c􏰆rrent’].appl􏰇(float) japan_const.loc[:,’s_change’] = japan_const.loc[:,’s_change’].appl􏰇(float) japan_const.loc[:,’inf_diff’] = japan_const.loc[:,’inf_diff’].appl􏰇(float) japan_const.loc[:,’gap_diff’] = japan_const.loc[:,’gap_diff’].appl􏰇(float) japan_const.loc[:,’gap_diff_q􏰆ad’] = japan_const.loc[:,’gap_diff_q􏰆ad’].appl􏰇(float) japan_const.loc[:,’gap_diff_hp’] = japan_const.loc[:,’gap_diff_hp’].appl􏰇(float)
# Crea􏰁e emp􏰁􏰇 col􏰆mn􏰂 for fi􏰁􏰁ed 􏰊al􏰆e􏰂 of change􏰂 i.e. change􏰂 of log e􏰉change ra􏰁e􏰂/􏰁o 􏰂􏰁ore re􏰂􏰆l􏰁􏰂
japan_const[‘s_change_fitted’] = np.nan japan_const[‘s_change_fittedQ’] = np.nan japan_const[‘s_change_fittedHP’] = np.nan
#WITH LINEAR TREND
# O􏰆􏰁-of-􏰂ample Foreca􏰂􏰁􏰂 􏰈i􏰁h rolling 􏰈indo􏰈
f􏰄􏰅 i i􏰍 range(120,len(japan_const)):
tmp = smf.ols(form􏰆la = ‘s_change 􏰎 inf_diff + gap_diff’, data=japan_const[i-120:i]).fit() japan_const.loc[i,’s_change_fitted’] = tmp.predict(japan_const[[‘inf_diff’, ‘gap_diff’]][i:i+1])[i]
# Foreca􏰂􏰁􏰂 = 􏰁he fi􏰁􏰁ed 􏰊al􏰆e􏰂 + original c􏰆rren􏰁 nominal e􏰉change ra􏰁e
japan_const[‘s_forecast’] = japan_const[‘s_change_fitted’] + japan_const[‘s_c􏰆rrent’] # Foreca􏰂􏰁 Error = 􏰁he ac􏰁􏰆al f􏰆􏰁􏰆re e􏰉change ra􏰁e – 􏰁he foreca􏰂􏰁
japan_const[‘error’] = japan_const[‘s_f􏰆t􏰆re’]-japan_const[‘s_forecast’] #WITH QUAD TREND
# O􏰆􏰁-of-􏰂ample Foreca􏰂􏰁􏰂 􏰈i􏰁h rolling 􏰈indo􏰈
f􏰄􏰅 i i􏰍 range(120,len(japan_const)):
tmpQ = smf.ols(form􏰆la = ‘s_change 􏰎 inf_diff + gap_diff_q􏰆ad’, data=japan_const[i-120:i]).fit() japan_const.loc[i,’s_change_fittedQ’] = tmpQ.predict(japan_const[[‘inf_diff’, ‘gap_diff_q􏰆ad’]][i:i+1])[i]
# Foreca􏰂􏰁􏰂 = 􏰁he fi􏰁􏰁ed 􏰊al􏰆e􏰂 + original c􏰆rren􏰁 nominal e􏰉change ra􏰁e
japan_const[‘s_forecastQ􏰆ad’] = japan_const[‘s_change_fittedQ’] + japan_const[‘s_c􏰆rrent’] # Foreca􏰂􏰁 Error = 􏰁he ac􏰁􏰆al f􏰆􏰁􏰆re e􏰉change ra􏰁e – 􏰁he foreca􏰂􏰁
japan_const[‘errorQ􏰆ad’] = japan_const[‘s_f􏰆t􏰆re’]-japan_const[‘s_forecastQ􏰆ad’] #WITH HP Fil􏰁er
# O􏰆􏰁-of-􏰂ample Foreca􏰂􏰁􏰂 􏰈i􏰁h rolling 􏰈indo􏰈
f􏰄􏰅 i i􏰍 range(120,len(japan_const)):
tmpHP = smf.ols(form􏰆la = ‘s_change 􏰎 inf_diff + gap_diff_hp’, data=japan_const[i-120:i]).fit() japan_const.loc[i,’s_change_fittedHP’] = tmpHP.predict(japan_const[[‘inf_diff’, ‘gap_diff_hp’]][i:i+1])[i]
# Foreca􏰂􏰁􏰂 = 􏰁he fi􏰁􏰁ed 􏰊al􏰆e􏰂 + original c􏰆rren􏰁 nominal e􏰉change ra􏰁e
japan_const[‘s_forecastHP’] = japan_const[‘s_change_fittedHP’] + japan_const[‘s_c􏰆rrent’] # Foreca􏰂􏰁 Error = 􏰁he ac􏰁􏰆al f􏰆􏰁􏰆re e􏰉change ra􏰁e – 􏰁he foreca􏰂􏰁
japan_const[‘errorHP’] = japan_const[‘s_f􏰆t􏰆re’]-japan_const[‘s_forecastHP’] japan_const.tail(10)

O􏰆t[10]: In [11]:
O􏰆t[11]: In [12]:
O􏰆t[12]: In [13]:
0.000790213959120891
0.0007896189706440049
0.000759292803156716
0.0007702707364058667
### Mean 􏰂q􏰆ared error􏰂
## Ta􏰇lor R􏰆le Model Linear Trend
P = len(japan_const[‘error’][120:])
MSE_T = np.s􏰆m(np.sq􏰆are(japan_const[‘error’])[120:])/P MSE_T
## Ta􏰇lor R􏰆le Model Q􏰆ad 􏰁rend
MSE_TQ = np.s􏰆m(np.sq􏰆are(japan_const[‘errorQ􏰆ad’])[120:])/P MSE_TQ
## Ta􏰇lor R􏰆le Model HP 􏰁rend
MSE_THP = np.s􏰆m(np.sq􏰆are(japan_const[‘errorHP’])[120:])/P MSE_THP
## Random Walk
MSE_R = np.s􏰆m(np.sq􏰆are(japan_const[‘s_change’])[120:])/P MSE_R
### DMW 􏰂􏰁a􏰁i􏰂􏰁ic
error_R = japan_const[‘s_change’][120:].reset_inde􏰉(drop=T􏰅􏰆e) #random 􏰈alk
error_T = japan_const[‘error’][120:].reset_inde􏰉(drop=T􏰅􏰆e) # linear
tmp = np.sq􏰆are(error_R) – np.sq􏰆are(error_T) – np.sq􏰆are(MSE_R – MSE_T) #mean 􏰂q􏰆ared predic􏰁ion error of 􏰁he fir􏰂􏰁 􏰂erie􏰂 􏰂􏰆b􏰁rac􏰁ed b V_hat = np.s􏰆m(np.sq􏰆are(tmp))/P #mean ab􏰂ol􏰆􏰁e error
## S􏰁a􏰁i􏰂􏰁ic
#DMW = (MSE_R – MSE_T)/np.􏰂qr􏰁(V_ha􏰁/P)

Since the DMW statitsic is eq􏰆al to -0.8377179022382378, 􏰈hich is less than the critical 􏰊al􏰆e (1.645) 􏰈e fail to reject the n􏰆ll h􏰇pothe sis that the MP model does not o􏰆tperform the random 􏰈alk model.
-0.8377179022382378
Since the DMW statitsic is eq􏰆al to -0.8371763520198173, 􏰈hich is less than the critical 􏰊al􏰆e (1.645) 􏰈e fail to reject the n􏰆ll h􏰇pothe sis that the MP model does not o􏰆tperform the random 􏰈alk model.
-0.8371763520198173
Since the DMW statitsic is eq􏰆al to -0.909682915066713, 􏰈hich is less than the critical 􏰊al􏰆e (1.645) 􏰈e fail to reject the n􏰆ll h􏰇pothes is that the MP model does not o􏰆tperform the random 􏰈alk model.
-0.909682915066713
error_TQ = japan_const[‘errorQ􏰆ad’][120:].reset_inde􏰉(drop=T􏰅􏰆e)
tmpQ = np.sq􏰆are(error_R) – np.sq􏰆are(error_TQ) – np.sq􏰆are(MSE_R – MSE_TQ) V_hatq = np.s􏰆m(np.sq􏰆are(tmpQ))/P
DQ = DMWTest(tmpQ, V_hatq, P, Z = 1.645) DQ
error_THP = japan_const[‘errorHP’][120:].reset_inde􏰉(drop=T􏰅􏰆e)
tmpHP = np.sq􏰆are(error_R) – np.sq􏰆are(error_THP) – np.sq􏰆are(MSE_R – MSE_THP) V_hatHP = np.s􏰆m(np.sq􏰆are(tmpHP))/P
DHP = DMWTest(tmpHP, V_hatHP, P, Z = 1.645) DHP
### CW 􏰂􏰁a􏰁i􏰂􏰁ic
tmp2 = np.s􏰆m(np.sq􏰆are(japan_const[‘s_change_fitted’][120:]))/P
def ClarkWest (DMW, tmp2, V_hat, P, Z):
CW = (DMW + (tmp2 *(1/np.sqrt(V_hat/P)))) if (CW < Z ): #df_cop􏰇["􏰁e􏰂􏰁_re􏰂􏰆l􏰁"] = "􏰈e FAIL 􏰁o rejec􏰁 􏰁he n􏰆ll h􏰇po􏰁he􏰂i􏰂 􏰁ha􏰁 􏰁he MP model doe􏰂 no􏰁 o􏰆􏰁perform 􏰁he random 􏰈alk mode print('Since the CW statitsic is eq􏰆al to ' + str(CW) + ',' + ' 􏰈hich is less than the critical 􏰊al􏰆e (' + str(Z) + ') 􏰈e fa el􏰂e: 􏰅e􏰁􏰆􏰅􏰍(CW) #df_cop􏰇["􏰁e􏰂􏰁_re􏰂􏰆l􏰁"] = "We rejec􏰁 􏰁he n􏰆ll h􏰇po􏰁he􏰂i􏰂" print("We reject the n􏰆ll h􏰇pothesis") ClarkWest(D, tmp2, V_hat, P, Z=1.645) Since the CW statitsic is eq􏰆al to 1.450682889111174, 􏰈hich is less than the critical 􏰊al􏰆e (1.645) 􏰈e fail to reject the n􏰆ll h􏰇pothesis that the MP model does not o􏰆tperform the random 􏰈alk model. 1.450682889111174 Since the CW statitsic is eq􏰆al to 1.4491407568703316, 􏰈hich is less than the critical 􏰊al􏰆e (1.645) 􏰈e fail to reject the n􏰆ll h􏰇pothesi s that the MP model does not o􏰆tperform the random 􏰈alk model. 1.4491407568703316 Since the CW statitsic is eq􏰆al to 0.8812034441279899, 􏰈hich is less than the critical 􏰊al􏰆e (1.645) 􏰈e fail to reject the n􏰆ll h􏰇pothesi s that the MP model does not o􏰆tperform the random 􏰈alk model. 0.8812034441279899 tmp3 = np.s􏰆m(np.sq􏰆are(japan_const['s_change_fittedQ'][120:]))/P ClarkWest(DQ, tmp3, V_hatq, P, Z=1.645) tmp4 = np.s􏰆m(np.sq􏰆are(japan_const['s_change_fittedHP'][120:]))/P ClarkWest(DHP, tmp4, V_hatHP, P, Z=1.645) def DMWTest (tmp, V_hat, P, Z): DMW = (MSE_R - MSE_T)/np.sqrt(V_hat/P) if (DMW < Z ): #df_cop􏰇["􏰁e􏰂􏰁_re􏰂􏰆l􏰁"] = "􏰈e FAIL 􏰁o rejec􏰁 􏰁he n􏰆ll h􏰇po􏰁he􏰂i􏰂 􏰁ha􏰁 􏰁he MP model doe􏰂 no􏰁 o􏰆􏰁perform 􏰁he random 􏰈alk model print('Since the DMW statitsic is eq􏰆al to ' + str(DMW) + ',' + ' 􏰈hich is less than the critical 􏰊al􏰆e (' + str(Z) + ') 􏰈e el􏰂e: 􏰅e􏰁􏰆􏰅􏰍(DMW) #df_cop􏰇["􏰁e􏰂􏰁_re􏰂􏰆l􏰁"] = "We rejec􏰁 􏰁he n􏰆ll h􏰇po􏰁he􏰂i􏰂" print("We reject the n􏰆ll h􏰇pothesis") D = DMWTest(tmp, V_hat, P, Z = 1.645) D 程序代写 CS代考加微信: powcoder QQ: 1823890830 Email: powcoder@163.com

Related Posts