def curveFromBin(binValues, pb, delta=1):
'''
binValues: 经排序的数值列表
'''
N = len(binValues)
binCumulateCount = np.array(range(1, N + 1))
li = Interpolate(binCumulateCount, binValues, delta)
cnt = pb * N / 100
bin = li.evalArray(cnt, sorted=True)
return CumulativeCurve(bin, cnt)
def merge(curves, pb):
"""
curves: 元素为(bins，counts)元组的列表
"""
binList = []
for c in curves:
binList.append(c[0])
bs = np.unique(np.ravel(np.array(binList)))
bsLen = len(bs)
binCounts = np.zeros(bsLen)
for cbin, cCount in curves:
li = Interpolate(cbin, cCount, 0)
cs = li.evalArray(bs, sorted=True)
binCounts = binCounts + cs
newTotal = binCounts[-1]
li = Interpolate(binCounts * 100 / newTotal, bs, 0)
newbin = li.evalArray(pb, sorted=True)
return CumulativeCurve(newbin, pb * newTotal / 100)
def curveFromBinCount(binCounts, pb, delta=1):
'''
binCounts: 元素为(bin，count)元组的列表，可以是多次累积，bin可以无序
'''
from collections import Counter
c0 = Counter()
for bin, cnt in binCounts:
c0[bin] += cnt
c1 = sorted(c0.items(), key=lambda item: item[0])
clen = len(c1)
if c1[0][1] == 0:
b = np.zeros(clen)
binCumulateCount = np.zeros(clen)
b[0] = c1[0][0]
binCumulateCount[0] = 0
for i in range(1, clen - 1):
b[i] = b[i - 1] + 2 * (c1[i][0] - b[i - 1])
binCumulateCount[i] = binCumulateCount[i - 1] + c1[i][1]
b[-1] = c1[-1][0]
binCumulateCount[-1] = binCumulateCount[-2]
else:
b = np.zeros(clen + 1)
binCumulateCount = np.zeros(clen + 1)
b[0] = c1[0][0] - delta
binCumulateCount[0] = 0
for i in range(clen):
b[i + 1] = c1[i][0]
binCumulateCount[i + 1] = binCumulateCount[i] + c1[i][1]
N = binCumulateCount[-1]
li = Interpolate(binCumulateCount, b, 0)
cnt = pb * N / 100
bin = li.evalArray(cnt, sorted=True)
return CumulativeCurve(bin, cnt)