def buc(df, pre, df2):
if df.shape[0] == 1:
binLst = [
bin(i)[3:] for i in range(2**df.shape[1], 2**(df.shape[1] + 1), 2)
]
for s in binLst:
tmpLst = [
'ALL' if s[i] == '1' else list(df.iloc[0])[i]
for i in range(len(s))
]
tmpLst = pre + tmpLst
df2.loc[len(df2)] = tmpLst
elif df.shape[1] == 1:
_pre = pre.copy()
_pre.append(sum(helper.project_data(df, 0)))
df2.loc[len(df2)] = _pre
else:
valLst = sorted(list(set(helper.project_data(df, 0).values)))
_pre = pre.copy()
for val in valLst:
subdf = helper.slice_data_dim0(df, val)
pre_ = _pre.copy()
pre_.append(val)
buc(subdf, pre_, df2)
subdf = helper.remove_first_dim(df)
pre_ = _pre.copy()
pre_.append('ALL')
buc(subdf, pre_, df2)
def buc_rec(input, l=[]):
dims = input.shape[1]
# Note that input is a DataFrame
if input.shape[0] == 1 and dims == 3:
a = input.iloc[0, 0]
b = input.iloc[0, 1]
s = input.iloc[0, 2]
ll.extend([
l + [a, b, s], l + [a, "ALL", s], l + ["ALL", b, s],
l + ["ALL", "ALL", s]
])
return
if dims == 1:
# only the measure dim
input_sum = sum(helper.project_data(input, 0))
input_sum = "%.1f" % input_sum
ll.append(l + [input_sum])
else:
# the general case
dim0_vals = set(helper.project_data(input, 0).values)
for dim0_v in dim0_vals:
sub_data = helper.slice_data_dim0(input, dim0_v)
buc_rec(sub_data, l + [dim0_v])
## for R_{ALL}
sub_data = helper.remove_first_dim(input)
buc_rec(sub_data, l + ['ALL'])
def _buc_rec_optimized(df, pre_num, df_out): # help function
# Note that input is a DataFrame
dims = df.shape[1]
if dims == 1:
# only the measure dim
input_sum = sum(helper.project_data(df, 0))
pre_num.append(input_sum)
df_out.loc[len(df_out)] = pre_num
else:
# the general case
dim0_vals = set(helper.project_data(df, 0).values)
temp_pre_num = deepcopy(pre_num)
for dim0_v in dim0_vals:
pre_num = deepcopy(temp_pre_num)
sub_data = helper.slice_data_dim0(df, dim0_v)
pre_num.append(dim0_v)
_buc_rec_optimized(sub_data, pre_num, df_out)
## for R_{ALL}
sub_data = helper.remove_first_dim(df)
pre_num = deepcopy(temp_pre_num)
pre_num.append("ALL")
_buc_rec_optimized(sub_data, pre_num, df_out)
def buc_rec_3_params(input, result, pre=[]):
rows = input.shape[0]
dims = input.shape[1]
new_pre = pre.copy()
if dims >= 2 and rows == 1:
new_pre += input.iloc[0, ].tolist()
new_rows = single_line_opt(tuple(new_pre), dims)
result += new_rows
return result
elif dims == 1:
# only the measure dim
input_sum = sum(helper.project_data(input, 0))
new_pre.append(input_sum)
result.append(new_pre)
return result
else:
# the general case
dim0_vals = set(helper.project_data(input, 0).values)
for dim0_v in dim0_vals:
new_pre.append(dim0_v)
sub_data = helper.slice_data_dim0(input, dim0_v)
result = buc_rec_3_params(sub_data, result, new_pre)
new_pre = pre.copy()
## for R_{ALL}
sub_data = helper.remove_first_dim(input)
new_pre.append('ALL')
result = buc_rec_3_params(sub_data, result, new_pre)
return result
def buc(df, result, prefix=[]):
dims = df.shape[1]
pre_cp = [i for i in prefix]
if dims == 1:
pre_cp.append(sum(helper.project_data(df, 0)))
result.append(pre_cp)
else:
dim0_vals = set(helper.project_data(df, 0).values)
for i in dim0_vals:
pre_cp.append(i)
rest_data = helper.slice_data_dim0(df, i)
buc(rest_data, result, pre_cp)
pre_cp = [i for i in prefix]
pre_cp.append('ALL')
rest_data = helper.remove_first_dim(df)
buc(rest_data, result, pre_cp)
return result
def my_buc_rec_optimized(df, pre, res): # help recursive function
dims = df.shape[1]
if df.shape[0] == 1:
single_tuple(df, pre, res)
elif dims == 1:
pre.append(sum(helper.project_data(df, 0)))
res.loc[len(res)] = pre
else:
vals = set(helper.project_data(df, 0).values)
pre_copy = pre.copy()
for val in vals:
pre = pre_copy.copy()
sub_data = helper.slice_data_dim0(df, val)
pre.append(val)
my_buc_rec_optimized(sub_data, pre, res)
sub_data = helper.remove_first_dim(df)
pre = pre_copy.copy()
pre.append("ALL")
my_buc_rec_optimized(sub_data, pre, res)
def cal_mul_tuple(df, result, check):
dims = df.shape[1]
check_trace = copy_list(check)
if dims == 1:
# only the measure dim
input_sum = sum(helper.project_data(df, 0))
check_trace.append(input_sum)
result.append(check_trace)
else:
# the general case
dim0_vals = set(helper.project_data(df, 0).values)
for dim0_v in dim0_vals:
check_trace.append(dim0_v)
sub_data = helper.slice_data_dim0(df, dim0_v)
cal_mul_tuple(sub_data, result, check_trace)
check_trace = copy_list(check)
## for R_{ALL}
sub_data = helper.remove_first_dim(df)
check_trace.append('ALL')
cal_mul_tuple(sub_data, result, check_trace)
return result
def buc_rec_general(df, result, prefix=[]):
dims = df.shape[1]
rows = df.shape[0]
pre_cp = [i for i in prefix]
if dims == 1:
input_sum = sum(helper.project_data(df, 0))
pre_cp.append(input_sum)
result.append(pre_cp)
elif rows == 1:
s_tuple = list(df.iloc[0])
tuples = buc_single(pre_cp, s_tuple, dims)
# tuples 是[[],[]....]形式
result += tuples
else:
dim0_vals = set(helper.project_data(df, 0).values)
for dim0_v in dim0_vals:
pre_cp.append(dim0_v)
sub_data = helper.slice_data_dim0(df, dim0_v)
buc_rec_general(sub_data, result, pre_cp)
pre_cp = [i for i in prefix]
pre_cp.append('ALL')
sub_data = helper.remove_first_dim(df)
buc_rec_general(sub_data, result, pre_cp)
return result
def buc(df, df_out, pre=[]):
dims = df.shape[1]
if dims == 1: # one column left
all = [] # sum result on column
for x in helper.project_data(df, 0):
all.append(int(x))
pre.append(sum(all))
df_out.loc[len(df_out)] = pre
pre.pop()
return 0
elif df.shape[0] == 1:
st_opt(df, df_out, pre)
else:
dim0_vals = set(helper.project_data(df, 0).values)
for dim0_v in sorted(dim0_vals):
sub_data = helper.slice_data_dim0(df, dim0_v)
pre.append(dim0_v)
buc(sub_data, df_out, pre)
pre.pop()
sub_data = helper.remove_first_dim(df)
pre.append("ALL")
buc(sub_data, df_out, pre)
pre.pop()
return 0
import pandas as pd