X_ldp['policy00'] = X_ldp['policy00'].replace([0, 3, 4, 5, 6, 7, 8, 9, 10],
[1, 2, 3, 3, 3, 4, 4, 5, 5])
X_dpj['policy00'] = X_dpj['policy00'].replace([0, 3, 4, 5, 6, 7, 8, 9, 10],
[1, 2, 3, 3, 3, 4, 4, 5, 5])
X_jrp['policy00'] = X_jrp['policy00'].replace([0, 3, 4, 5, 6, 7, 8, 9, 10],
[1, 2, 3, 3, 3, 4, 4, 5, 5])
X = pd.concat([X_ldp, X_dpj, X_jrp])
print(X_ldp.shape, X_dpj.shape, X_jrp.shape, X.shape)
##Disctionay for Level and Party
party = {"LDP": "LDP", "DPJ": "DPJ", "JRP": "JRP"}
##Fitting cMCA and export plots
cmca = CMCA(n_components=2, copy=True, check_input=True)
cmca = cmca.fit(fg=X_ldp.iloc[:, 6:X.shape[1]],
bg=X_dpj.iloc[:, 6:X.shape[1]],
alpha=1.5)
Y_fg = np.array(cmca.transform(X_ldp.iloc[:, 6:X.shape[1]]))
Y_bg = np.array(cmca.transform(X_dpj.iloc[:, 6:X.shape[1]]))
f = plt.figure()
plt.xlim([-1.5, 2])
plt.ylim([-1.5, 1.5])
plt.scatter(Y_fg[:, 0],
Y_fg[:, 1],
c=tableau10[X_ldp["psup_short"].iloc[0]],
label=party[X_ldp["psup_short"].iloc[0]],
alpha=0.8,
linewidths=0)
plt.scatter(Y_bg[:, 0],
Y_bg[:, 1],
fillna_based_on_dtype(X_r)
return (X_d, X_r)
X_d, X_r = csv_to_mats_2('./issuevalue_short.csv')
X_d['partyid'] = X_d['partyid'].replace([3, 2], 1)
X_r['partyid'] = X_r['partyid'].replace([5, 6, 7], 2)
X = pd.concat([X_d, X_r])
##Disctionay for Level and Party
party = {1: "Dem", 2: "Rep"}
cmca = CMCA(n_components=2, copy=True, check_input=True)
cmca = cmca.fit(fg=X_r.iloc[:, 1:(X.shape[1] - 1)],
bg=X_d.iloc[:, 1:(X.shape[1] - 1)],
alpha=1.5)
Y_fg = np.array(cmca.transform(X_r.iloc[:, 1:(X.shape[1] - 1)]))
Y_bg = np.array(cmca.transform(X_d.iloc[:, 1:(X.shape[1] - 1)]))
Y_fg_col = np.array(cmca.transform(X_r.iloc[:, 1:(X.shape[1] - 1)],
axis='col'))
prefix_to_info = cmca.gen_prefix_to_info()
used_others_label = False
f = plt.figure()
for key in prefix_to_info.keys():
indices = prefix_to_info[key]['indices']
rank = prefix_to_info[key]['loading_ranks_norm_1']
rank = rank if rank < 9 else -1
texts = [
int(float(postfix)) for postfix in prefix_to_info[key]['postfixes']
##Disctionay for Level and Party
party = {
1: "Con",
2: "Lab",
3: "LD",
4: "SNP",
5: "Green",
6: "UKIP",
7: "Other"
}
##Fitting cMCA and export plots
cmca = CMCA(n_components=2, copy=True, check_input=True)
cmca = cmca.fit(fg=X_con.iloc[:, 0:(X_con.shape[1] - 3)],
bg=X_uip.iloc[:, 0:(X_uip.shape[1] - 3)],
alpha=100)
Y_fg = np.array(cmca.transform(X_con.iloc[:, 0:(X.shape[1] - 3)]))
Y_bg = np.array(cmca.transform(X_uip.iloc[:, 0:(X.shape[1] - 3)]))
Y_fg_col = np.array(
cmca.transform(X_con.iloc[:, 0:(X.shape[1] - 3)], axis='col'))
prefix_to_info = cmca.gen_prefix_to_info()
used_others_label = False
f = plt.figure()
for key in prefix_to_info.keys():
indices = prefix_to_info[key]['indices']
rank_1 = prefix_to_info[key]['loading_ranks_norm_0']
rank_1 = rank_1 if rank_1 < 9 else -1
texts = [
y = np.array(df.iloc[:, 0])
fg = X.iloc[y == 'democrat']
bg = X.iloc[y == 'republican']
# alpha = 0 (normal MCA on fg)
# alpha = 10 (contrastive MCA fg vs bg)
# alpha = 'auto' (contrastive MCA with auto selection of alpha)
cmca = CMCA(n_components=2)
for alpha in [0, 10, 'auto']:
### cMCA
auto_alpha = False
if alpha == 'auto':
alpha = None
auto_alpha = True
cmca.fit(fg, bg, alpha=alpha, auto_alpha_selection=auto_alpha)
# row coordinates (cloud of individuals)
Y_fg_row = np.array(cmca.transform(fg, axis='row'))
Y_bg_row = np.array(cmca.transform(bg, axis='row'))
# col coordinates (cloud of categories)
Y_fg_col = np.array(cmca.transform(fg, axis='col'))
Y_bg_col = np.array(cmca.transform(bg, axis='col'))
# cPC loadings
loadings = cmca.loadings
# category names
categories = cmca.categories
##Disctionay for Level and Party
party = {
1: "Con",
2: "Lab",
3: "LD",
4: "SNP",
5: "Green",
6: "UKIP",
7: "Other"
}
##Fitting cMCA and export plots
cmca = CMCA(n_components=2, copy=True, check_input=True)
cmca = cmca.fit(fg=X_lab.iloc[:, 0:(X_lab.shape[1] - 3)],
bg=X_uip.iloc[:, 0:(X_uip.shape[1] - 3)],
alpha=10)
Y_fg = np.array(cmca.transform(X_lab.iloc[:, 0:(X.shape[1] - 3)]))
Y_bg = np.array(cmca.transform(X_uip.iloc[:, 0:(X.shape[1] - 3)]))
Y_fg_col = np.array(
cmca.transform(X_lab.iloc[:, 0:(X.shape[1] - 3)], axis='col'))
prefix_to_info = cmca.gen_prefix_to_info()
used_others_label = False
f = plt.figure()
for key in prefix_to_info.keys():
indices = prefix_to_info[key]['indices']
rank = prefix_to_info[key]['loading_ranks_norm_1']
rank = rank if rank < 9 else -1
texts = [
fillna_based_on_dtype(X_uip)
fillna_based_on_dtype(X_oth)
return(X_con, X_lab, X_ldp, X_snp, X_gre, X_uip, X_oth)
X_con, X_lab, X_ldp, X_snp, X_gre, X_uip, X_oth = df_to_mat(df)
X = pd.concat([X_con, X_lab, X_ldp, X_snp, X_gre, X_uip, X_oth])
print(X_con.shape, X_lab.shape, X_ldp.shape, X_snp.shape, X_gre.shape, X_uip.shape, X_oth.shape, X.shape)
##Disctionay for Level and Party
party = {1:"Con", 2:"Lab", 3:"LD", 4:"SNP", 5:"Green", 6:"UKIP", 7:"Other"}
##Fitting cMCA and export plots
cmca = CMCA(n_components=2, copy=True, check_input=True)
cmca = cmca.fit(fg=X_lab.iloc[:,0:(X_lab.shape[1]-3)], bg=X_con.iloc[:,0:(X_con.shape[1]-3)], alpha=1.5)
Y_fg = np.array(cmca.transform(X_lab.iloc[:,0:(X.shape[1]-3)]))
Y_bg = np.array(cmca.transform(X_con.iloc[:,0:(X.shape[1]-3)]))
Y_fg_col = np.array(cmca.transform(X_lab.iloc[:,0:(X.shape[1]-3)], axis='col'))
prefix_to_info = cmca.gen_prefix_to_info()
f_6 = plt.figure()
plt.xlim([-2.5, 2.5])
plt.ylim([-2.5, 2.5])
plt.scatter(Y_fg[:, 0], Y_fg[:, 1], c=tableau10[X_lab["prtclcgb"].iloc[0]], label=party[X_lab["prtclcgb"].iloc[0]], alpha=0.3, linewidths=0)
plt.scatter(Y_bg[:, 0], Y_bg[:, 1], c=tableau10[X_con["prtclcgb"].iloc[0]], label=party[X_con["prtclcgb"].iloc[0]], alpha=0.3, linewidths=0)
handles, labels = plt.gca().get_legend_handles_labels()
handles = [handles[1],handles[0]]
labels = ["Con","Lab"]
plt.legend(handles, labels, loc="lower right", shadow=False, scatterpoints=1, fontsize=8)