def _start_r_interpreter(self):
'''Find and start a R intepreter
On windows systems use the packaged R env.
On Posix systems use the system installed R env.
Will raise RNotFoundException
'''
if sys.platform.startswith('win32'):
'''A MS Windows kind of system'''
logger.info("Running on a MS Windows system")
Rwbin = op.join(self.r_origo, 'R-3.3.1', 'bin', 'R.exe')
Rlib = op.join(self.r_origo, 'R-3.3.1', 'library')
logger.info("Try Windows R path: {0}".format(Rwbin))
if op.exists(Rwbin):
logger.info("R.exe found")
self.r = pyper.R(RCMD=Rwbin, use_pandas=True)
self.r('.libPaths("{0}")'.format(Rlib))
else:
'''Try to run R from system path
Give message if R is not found
'''
self.r = pyper.R(use_pandas=True)
# raise RNotFoundException()
else:
'''Not Windows, assumed to be a POSIX system
OS X Darwin or a Linux flavor
'''
logger.info("Assumed to run on a POSIX system")
Rxbin = self._find_posix_system_R()
logger.info("System R found at path: {0}".format(Rxbin))
self.r = pyper.R(RCMD=Rxbin, use_pandas=True)
def surv_ci(data, pred_col, duration_col, event_col):
"""Concordance Index
Parameters
----------
data : pandas.DataFrame
Full survival data.
pred_col : str
Name of column indicating log hazard ratio.
duration_col : str
Name of column indicating time.
event_col : str
Name of column indicating event.
Returns
-------
`dict`
Object of dict include details about CI.
Examples
--------
>>> surv_ci(data, 'Pred', 'T', 'E')
"""
X = data[pred_col].values
T = data[duration_col].values
E = data[event_col].values
r = pr.R(use_pandas=True)
r("library('survival')")
r("library('Hmisc')")
r.assign("t", T)
r.assign("e", E)
r.assign("x", X)
r("src <- rcorr.cens(-x, Surv(t, e))")
return r.src
def get_dict_correction(set_XC, designed_barcode,
metric="seqlev", distance="2"):
if type(distance) not in (int, long):
raise InvalidArgumentError
if distance not in (0, 1, 2):
print("distance must be 0, 1, or 2")
raise InvalidArgumentError
r = pr.R()
r("library(DNABarcodes)")
r.assign("list_XC", list(set_XC))
r.assign("designed_barcode", designed_barcode)
if metric == "seqlev":
r("demultiplexed <- demultiplex(list_XC, designed_barcode, metric='seqlev')")
elif metric == "hamming":
r("demultiplexed <- demultiplex(list_XC, designed_barcode, metric='hamming')")
else:
print("metric must be 'seqlev' or 'hamming'")
raise InvalidArgumentError
df_correction = r.get("demultiplexed")
df_correction.columns = [x.replace(" ", "") for x in df_correction.columns]
df_correction_filt = (df_correction[df_correction.distance <= distance]
[['read', 'barcode']])
dict_correct = df_correction_filt.set_index('read').to_dict()['barcode']
return dict_correct
def loess_res(ts, bass = 0):
'''
Fit loess curve using Friedman's Super Smoother and return residuals
Parameters
----------
ts: pandas series
time series of values
bass: int
smoothing parameter of curve. values up to 10 for more smoothness
Returns
-------
list
series of residuals from loess curve fit
'''
ts = ts.tolist()
# create a R instance
r = pr.R(use_pandas = True)
# pass ts from python to R as Y, and pass bass parameter
r.assign("Y", ts)
r.assign("bass", bass)
# fit friedman's super smoother on ts and extract the fitted values
r("fit = supsmu(x=1:length(Y), y=Y, bass=bass)$y")
# pass fitted values from r to python
fit = r.get("fit")
# residuals from loess fit
residuals = ts - fit
return(residuals.tolist())
def icar(file_path, algo, n_components, wl_range_start, wl_range_end):
data = pd.read_csv(file_path, index_col=0).T
data = data.loc[:, wl_range_start:wl_range_end].T
wl = data.index.values.astype('float32')
# Rのインスタンスを作る
r = pyper.R(use_pandas='True')
# PythonのオブジェクトをRに渡す
r.assign('data', data)
r.assign('wl', wl)
r.assign('n_components', n_components)
# Rのコードを実行する
r("library(ica)")
r("X <- data")
if algo == 'FastICA':
r("a <- icafast(X, n_components)")
elif algo == 'InfoMax':
r("a <- icaimax(X, n_components)")
else:
r("a <- icajade(X, n_components)")
r("ics <- cbind(wl, a$S)")
# Pythonでrのオブジェクトを読む
ics = r.get("ics")
ics_df = pd.DataFrame(ics)
return ics_df
def readModelFile(modelFileName, jobName):
if os.path.isfile(modelFileName):
r = pr.R()
r(IMPORT_R_WRAPPER)
r.assign('params', json.dumps({'rdsFile': modelFileName}))
rOutput = r('model <- readFromRDS(params)')
log.info(rOutput)
results = r.get('model')
del r
rOutput = None
if results:
model = json.loads(results)
maxTimePoint = model['maxTimePoint'][0]
covariates = model['covariates']
if (len(model['jobName'])):
jobName = model['jobName'][0]
return {
'jobName': jobName,
'maxTimePoint': maxTimePoint,
'interceptOnly': True if len(covariates) == 0 else False
}
else:
message = "Couldn't read Time Points from RDS file!"
log.error(message)
return message
else:
message = "Model file does not exist!"
log.error(message)
return message
def __init__(self,
training_marginal_dist_matrix: np.matrix,
family: str,
param=None,
dim=None):
if family not in ['gumbel', 'clayton', 'frank', 'normal', 'indep']:
print('Copula family "' + family + '" is not supported.')
raise ValueError
if training_marginal_dist_matrix.size == 0:
if dim is None:
raise ValueError
self._dimension = dim
else:
self._dimension = training_marginal_dist_matrix.shape[1]
self._r_engine = pyper.R()
self._r_engine.assign("py.training.marginal.dist.matrix",
training_marginal_dist_matrix)
self._r_engine.assign("py.cop.name", family)
self._r_engine.assign("py.param", param)
self._r_engine.assign("py.dim", self._dimension)
self._r_engine('source("copula/copula.R")')
self._r_engine('source("copula/copula.R")')
trained_param = self._r_engine.get("trained.param")
self.trained_param = trained_param #asahi
if trained_param is None:
self._r_engine('trained <- indepCopula(dim=%d)' % self._dimension)
print('indep')
else:
print(trained_param)
def arimaTemp(filename):
processTemp()
predictions = []
r = pr.R(RCMD="C:\\Program Files\\R\\R-3.1.2\\bin\\R", use_numpy=True, use_pandas=True)
datetimes = np.arange('2008-07-01 00:00:00','2008-07-08 00:00:00',dtype='datetime64[h]')
for j in range(1,12):
data = pd.read_csv('../data/outputs/temp_history_processed_station_%s.csv'%j, parse_dates='datetime')
subts = data["value"]
print 'Predictions for zone %s'%j
results = arima(subts,r)
results = pd.DataFrame(results, columns=['value'])
results['datetime'] = datetimes
results['station_id'] = j
predictions.append(results)
concatPredictions = pd.concat(predictions)
concatPredictions.to_csv(filename, index=False, date_format='%Y-%m-%d %H:%M:%S', mode='a')
def plot_scatter_polya(in_fisher, out_png):
"""plot a scatter plot of the expression of polya sites"""
lhs, rhs = in_fisher.split('vs.')
lhs, rhs = short_name(lhs), short_name(rhs)
try:
compare_type = in_fisher.replace('.fisher_test', '').split('polya.')[1]
except:
compare_type = 'ALL'
R_script = r"""
library(lattice)
d<-read.table(file="%(in_fisher)s", header=TRUE, sep="\t")
png('%(out_png)s')
exp1 <- c(d$exp1_upstream_count, d$exp1_downstream_count)
exp2 <- c(d$exp2_upstream_count, d$exp2_downstream_count)
sig_sites <- d$fisher_p_two_sided < .05
exp1_sig = c(d$exp1_upstream_count[sig_sites], d$exp1_downstream_count[sig_sites])
exp2_sig = c(d$exp2_upstream_count[sig_sites], d$exp2_downstream_count[sig_sites])
plot(log2(exp1), log2(exp2), cex=.8, col='lightgray', pch=20, xlab="%(xlab)s", ylab="%(ylab)s", main="All sites for %(lhs)s vs. %(rhs)s in %(compare_type)s", sub=paste("R^2 is ", cor(exp1, exp2)))
dev.off()
""" % dict(plot_label=r'Poly-A for\n%s' % in_fisher,
xlab="log2(%s)" % (lhs),
ylab="log2(%s)" % (rhs),
in_fisher=in_fisher, out_png=out_png, lhs=lhs, rhs=rhs,
compare_type=compare_type)
# print R_script
r = pyper.R()
r(R_script)
def plot_differential_polya(in_fisher, out_pattern, out_template):
"""plot a scatter plot of the log-expression difference of polya sites"""
lhs, rhs = in_fisher.split('vs.')
lhs, rhs = short_name(lhs), short_name(rhs)
try:
compare_type = in_fisher.replace('.fisher_test', '').split('polya.')[1]
except:
compare_type = 'ALL'
for max_pval in [.05, .01, .001]:
out_png = out_template % ('pval_%s' % max_pval)
R_script = r"""
library(lattice)
d<-read.table(file="%(in_fisher)s", header=TRUE, sep="\t")
png('%(out_png)s')
sig_sites <- d$fisher_p_two_sided < %(max_pval)s
exp1_proximal = d$exp1_upstream_count[sig_sites]
exp1_distal = d$exp1_downstream_count[sig_sites]
exp2_proximal = d$exp2_upstream_count[sig_sites]
exp2_distal = d$exp2_downstream_count[sig_sites]
plot(log2(d$exp1_upstream_count/d$exp2_upstream_count), log2(d$exp1_downstream_count/d$exp2_downstream_count), cex=.8, col='lightgray', pch=20, xlab="%(xlab)s", ylab="%(ylab)s", main="Significant sites for %(lhs)s vs. %(rhs)s in %(compare_type)s", sub=paste("Significant sites:", sum(sig_sites), "/", dim(d)[1]))
points(log2(exp1_proximal/exp2_proximal), log2(exp1_distal/exp2_distal), col='red', cex=.8, pch=20)
dev.off()
""" % dict(plot_label=r'Differential Poly-A for\n%s' % in_fisher,
xlab="log2(%s/%s)-proximal" % (lhs, rhs),
ylab="log2(%s/%s)-distal" % (lhs, rhs),
in_fisher=in_fisher, out_png=out_png, lhs=lhs, rhs=rhs,
compare_type=compare_type, max_pval=max_pval)
# print R_script
r = pyper.R()
r(R_script)
def on_get(self, req, resp):
r = pr.R()
# https://www.r-tutor.com/elementary-statistics/quantitative-data/frequency-distribution-quantitative-data
r("duration = faithful$eruptions")
r("breaks = seq(1.5, 5.5, by=0.5)")
r("duration.cut = cut(duration, breaks, right=FALSE)")
r("duration.freq = table(duration.cut)")
resp.body = r("duration.freq")
def pc_rlib(d_dt, threshold, skel_method, verbose):
import pandas
import pyper
if skel_method == "default":
method = "original"
else:
method = skel_method
input_data = d_dt
#input_data = {}
#for nid, ns in nsdict.iteritems():
# input_data[nid] = ns.get_values()
r = pyper.R(use_pandas='True')
r("library(pcalg)")
r("library(graph)")
df = pandas.DataFrame(input_data)
r.assign("input.df", df)
r.assign("method", method)
r("evts = as.matrix(input.df)")
#print r("evts")
#r("t(evts)")
#r("save(evts, file='rtemp')")
r.assign("event.num", len(input_data))
r.assign("threshold", threshold)
r.assign("verbose.flag", verbose)
print r("""
pc.result <- pc(suffStat = list(dm = evts, adaptDF = FALSE),
indepTest = binCItest, alpha = threshold, skel.method = method,
labels = as.character(seq(event.num)-1), verbose = verbose.flag)
""")
#print r("""
# pc.result <- pc(suffStat = list(dm = evts, adaptDF = FALSE),
# indepTest = binCItest, alpha = threshold,
# labels = as.character(seq(event.num)-1), verbose = TRUE)
#""")
r("node.num <- length(nodes(pc.result@graph))")
g = nx.DiGraph()
for i in range(r.get("node.num")):
r.assign("i", i)
edges = r.get("pc.result@graph@edgeL[[as.character(i)]]$edges")
if edges is None:
pass
elif type(edges) == int:
g.add_edge(i, edges - 1)
elif type(edges) == np.ndarray:
for edge in edges:
g.add_edge(i, edge - 1)
else:
raise ValueError("edges is unknown type {0}".format(type(edges)))
return g
def ping():
try:
r = pr.R()
return r['"true"']
except Exception as e:
log.exception("Exception occurred")
return buildFailure({
"status": False,
"statusMessage": "Call R failed!"
})
def RdataToHDF5(fileName, variableName, path=None):
r = pyper.R()
if path is not None:
r["setwd('%s')" % path]
r['load("%s")' % fileName]
r['library(rhdf5)']
try:
r['h5createFile("%s.h5")' % fileName]
except pyper.RError:
pass # typically this is because the file already exists
# TODO: determine if something else went wrong
r['h5write(%s, "%s.h5","%s")' % (variableName, fileName, variableName)]
def make_basic_colormap(todir: str,
width: int,
height: int,
encoding="CP932") -> None:
'''shapeファイルから行政地区単位でカラーマップを作成
行政地区単位で異なる色で配色される．
しかしあくまで色と行政地区が対応しているのみで
行政地区とその名前・コードとの対応はなされない．
Parameters
----------
todir : str
shapeファイルまでのディレクトリ
width : int
出力svgファイルの幅
height : int
出力svgファイルの高さ
encoding : str
入力svgファイルに使用されている文字コード
'''
#create R object
r = pyper.R()
r("library(sf)")
r("library(ggplot2)")
#input shapefile
shapefile = (glob.glob(os.path.join(todir, "*.shp")))[0]
r.assign('shapefile', shapefile)
#make tmporary directory.
os.makedirs('./tmp', exist_ok=True)
#set output svg file name
svgfile = './tmp/tmp.svg'
r.assign('svgfile', svgfile)
#output OsakaMap
#Each area are colored according to municipality code.
r.assign('param1', width)
r.assign('param2', height)
encoding_to_r = "ENCODING=" + encoding + '"'
r.assign('option', encoding_to_r)
r('shp <- sf::st_read(shapefile,options=option)')
r('svg(svgfile, width=param1, height=param2)')
r('ggplot()+geom_sf(data=shp,aes(fill=N03_007))')
r('dev.off()')
def __pipeData(self):
'''
This function pipes data to R environment and declares R variables
NOTE: R must be installed in this script's environment
'''
try:
self.r = pyper.R(use_numpy = True) # For data input as numpy array
self.r("chooseCRANmirror(ind=10)") # Choose Canadian host as R mirror (for package downloads)
self.r.assign('data', self.data)
except:
print 'There was an error piping data to R environment. Please ensure that R is properly installed and all PATH variables are correct.'
def plot_balls(balls, depth):
global save_dir
r = pyper.R()
r_code = '''
par(pty="s")
png("{0}/depth_{1}.png")
data <- read.csv('{0}/depth_{1}.csv', header = FALSE)
p = data$V1
q = data$V2
col1 <- densCols(p, q, colramp = colorRampPalette(c("white", "orange", "red")))
plot(0, 0, type = "n", xlim = c(0, 100), ylim = c(0, 100),xlab = "x", ylab = "y")
points(p, q, col = col1, pch = 3)
dev.off()
'''.format(save_dir, depth)
r(r_code)
def draw_expression_correlation(in_data, out_png):
"""Correlation test to see if the correlation between expression values and
peak quality (score column from peak file).
"""
R_script = r"""
png('%(out_png)s')
d<-read.table(file="%(in_data)s", header=TRUE, sep="\t");
library(lattice);
r <- cor.test(d$expression_val, d$peak_score)
plot(d$expression_val, d$peak_score, xlab="expression value", ylab="peak score")
title(paste("R^2 = ", r$estimate, ", p-value = ", r$p.value));
dev.off()
""" % dict(in_data=in_data, out_png=out_png)
#print R_script
r = pyper.R()
r(R_script)
def youden_onecut(data, pred_col, duration_col, event_col, pt=None):
"""Cutoff maximize Youden Index.
Parameters
----------
data : pandas.DataFrame
full survival data.
pred_col : str
Name of column to reference for dividing groups.
duration_col : str
Name of column indicating time.
event_col : str
Name of column indicating event.
pt : int, default None
Predicted time.
Returns
-------
float
Value indicating cutoff for pred_col of data.
Examples
--------
>>> youden_onecut(data, 'X', 'T', 'E')
"""
X = data[pred_col].values
T = data[duration_col].values
E = data[event_col].values
if pt is None:
pt = T.max()
r = pr.R(use_pandas=True)
r.assign("t", T)
r.assign("e", E)
r.assign("mkr", np.reshape(X, E.shape))
r.assign("pt", pt)
r.assign("mtd", "KM")
r.assign("nobs", X.shape[0])
r("library(survivalROC)")
r("src <- survivalROC(Stime = t, status = e, marker = mkr, predict.time = pt, span = 0.25*nobs^(-0.20))")
r("Youden <- src$TP-src$FP")
r("cutoff <- src$cut.values[which(Youden == max(Youden), arr.ind = T)]")
r("abline(0,1)")
return r.cutoff
def exec_r(self):
import pyper
r = pyper.R(use_pandas='True')
r("suppressWarnings(require(tseries,warn.conflicts = FALSE,quietly=TRUE))"
)
#r("source(file='mswm.R')")#, encoding='utf-8'
#r("result<-m.lm$coefficients[2]")
#r("result1<-y1")
#r("result2<-y2")
#print(r.get("result"))
#print("completed")
for i in range(self._term, self._return_df.shape[0]):
self._return_df.iloc[i - self._term:i].to_csv('run.csv')
r("df <- read.csv('run.csv', header = T)")
r("d <- diff(df$Last)")
r("x <- factor(sign(d[-which(d %in% 0)]))")
r("run_result <- runs.test(x)")
r("p_value <- run_rsult$p.value")
print(r.get('p_value'))
def fit_predict(self, X_train, y_train, X_test, y_test):
X = np.vstack((X_train, X_test))
df = DataFrame(X)
r = pr.R(use_pandas=True)
r.assign("X", df)
# r('print(X)') # if I remove this line pyper get stuck.....
install_dir = os.path.dirname(
os.path.abspath(inspect.getfile(
inspect.currentframe()))) # script directory
r('source("' + install_dir + '/RF.R")')
r('set.seed(0)')
r('no.forests=' + str(int(self.nforests)))
r('no.trees=' + str(int(self.ntree)))
r('rfdist <- RFdist(X, mtry1=3, no.trees, no.forests, '
'addcl1=T, addcl2=F, imp=T, oob.prox1=T)')
r('labelRF=outlier(rfdist$cl1)')
return -np.array(r.get('labelRF'))[X_train.shape[0]:]
def surv_roc(data, pred_col, duration_col, event_col, pt=None):
"""Get survival ROC at predicted time.
Parameters
----------
data : pandas.DataFrame
Full survival data.
pred_col : str
Name of column to reference for dividing groups.
duration_col : str
Name of column indicating time.
event_col : str
Name of column indicating event.
pt : int
Predicted time.
Returns
-------
`dict`
Object of dict include "FP", "TP" and "AUC" in ROC.
Examples
--------
>>> surv_roc(data, 'X', 'T', 'E', pt=5)
"""
X = data[pred_col].values
T = data[duration_col].values
E = data[event_col].values
if pt is None:
pt = T.max()
r = pr.R(use_pandas=True)
r.assign("t", T)
r.assign("e", E)
r.assign("mkr", np.reshape(X, E.shape))
r.assign("pt", pt)
r.assign("mtd", "KM")
r.assign("nobs", X.shape[0])
# different predict.time may plot 1, 5, or 10 year ROC
r("src<-survivalROC::survivalROC(Stime = t, status = e, marker = mkr, predict.time = pt, span = 0.25*nobs^(-0.20))")
# r.src['AUC'] r.src['FP'], r.src['TP']
return r.src
def get_quantile_errors(data, quantiles):
"""
Gets arbitrary quantile values for given data
This function uses the PypeR library (download from
http://sourceforge.net/projects/rinpy/) to call commands from the R
language used to calculate an arbitrary number of quantile intervals of
a numpy array. In the case of bi-dimensional numpy arrays, it will
calculate these statistics along the first axis, meaning that each line
corresponds to an independent data set.
Input:
- data numpy.ndarray : data input, each line being a data set
(ndim=2,dtype=float)
- quantiles numpy.ndarray : percentages (ndim=1,dtype=float)
Ex: quartiles : array([0.25,0.5,0.75])
Output:
- qtl_errors numpy.ndarray : quantile error values (ndim=2,dtype=float)
---
"""
myR = pyper.R()
myR['data'] = np.transpose(data)
myR['quantiles'] = quantiles
# Calculate the quantiles for each data line
myR("""qtls <- t(sapply(as.data.frame(data),function(x) quantile(x,
quantiles,names=FALSE)))""")
qtl_errors = myR['qtls']
del myR
return qtl_errors
def plot_nearest_features(in_distances, out_png, test_out, window_size=20):
"""Plot a density of the distance to the nearest features"""
print out_png
print test_out
R_script = r"""
png('%(out_png)s')
d<-read.table(file="%(in_data)s", header=TRUE, sep="\t");
d = d / 1000;
library(lattice);
plot(density(unlist(d[1])[unlist(d[1]) < %(window_size)s & unlist(d[1]) > -%(window_size)s], na.rm=TRUE), main="Feature densities around peaks", xlab="Distance (kb)", ylab="Density", xlim=c(-%(window_size)s,%(window_size)s))
index = 1
r = rainbow(length(d))
for (i in d) {
i = i[i < %(window_size)s & i > -%(window_size)s]
lines(density(i, from=-%(window_size)s, to=%(window_size)s, na.rm=TRUE), col=r[index])
index = index + 1
}
legend("topleft", legend=names(d), col=r, lty=1)
dev.off()
""" % dict(in_data=in_distances, out_png=out_png, window_size=window_size)
print R_script
r = pyper.R()
r(R_script)
def plot_ttest_polya(in_ttest, out_png):
"""plot the t-test averages used as a scatter plot of the expression of polya sites"""
lhs, rhs = in_ttest.split('vs.')
lhs, rhs = short_name(lhs), short_name(rhs)
try:
compare_type = in_ttest.replace('.t_test', '').split('polya.')[1]
except:
compare_type = 'ALL'
R_script = r"""
library(lattice)
d<-read.table(file="%(in_ttest)s", header=TRUE, sep="\t")
png('%(out_png)s')
exp1 <- unlist(lapply(lapply(strsplit(gsub("]", "", gsub("[", "", d$exp1_count, fixed=TRUE), fixed=TRUE), ", ", fixed=TRUE), as.numeric), mean))
exp2 <- unlist(lapply(lapply(strsplit(gsub("]", "", gsub("[", "", d$exp2_count, fixed=TRUE), fixed=TRUE), ", ", fixed=TRUE), as.numeric), mean))
sig_sites <- d$ttest_pvalue < .05
# upregulated means t-statistic is positive => exp1 < exp2
exp1_bigger <- d$ttest_pvalue < .05 & d$ttest_stat > 0
exp2_bigger <- d$ttest_pvalue < .05 & d$ttest_stat < 0
exp1_sig = exp1[sig_sites]
exp2_sig = exp2[sig_sites]
plot(log2(exp1), log2(exp2), cex=.8, col='lightgray', pch=20, xlab="", ylab="%(ylab)s", main="All sites for %(lhs)s vs. %(rhs)s in %(compare_type)s", sub=paste("R^2 is ", cor(exp1, exp2),"\nSig sites x > y: ", sum(exp1_bigger), "\nSig sites x < y: ", sum(exp2_bigger)))
points(log2(exp1_sig), log2(exp2_sig), col='red', cex=.8, pch=20)
dev.off()
""" % dict(plot_label=r'Poly-A for\n%s' % in_ttest,
xlab="log2(%s)" % (lhs),
ylab="log2(%s)" % (rhs),
in_ttest=in_ttest, out_png=out_png, lhs=lhs, rhs=rhs,
compare_type=compare_type)
# print R_script
r = pyper.R()
r(R_script)
def post(self):
#データベース
db = pd.read_csv("data.csv")
params = pd.read_csv("params.csv")
#入力されたデータ(値は0~5の6段階評価)
user_data = pd.DataFrame([[int(self.get_argument("q1")),
int(self.get_argument("q2")),int(self.get_argument("q3")),int(self.get_argument("q4")),int(self.get_argument("q5")),int(self.get_argument("q6")),int(self.get_argument("q7")),int(self.get_argument("q8")),int(self.get_argument("q9")),int(self.get_argument("q10")),int(self.get_argument("q11")),int(self.get_argument("q12")),int(self.get_argument("q13")),int(self.get_argument("q14")),int(self.get_argument("q15")),int(self.get_argument("q16")),"偏差値"]],
columns=["V1","V2","V3","V4","V5","V6","V7","V8","V9","V10","V11","V12","V13","V14","V15","V16","偏差値"])
data = (db.append(user_data, ignore_index=True))
#Rでの分析
r = pyper.R(use_pandas='True')
r.assign("data", data)
r.assign("params", params)
r("source('myfunc/myfunc.R')")
r("library(psych)")
r("library(irtoys)")
r("library(ltm)")
#17列目の偏差値は分析データから外し、最新行のみ分析
r("data<-data[nrow(data),1:16]")
#母数の推定
r("a <- grm.theta(data,a=params[,6]/1.7,bc=params[,c(1,2,3,4,5)],D=1.7,method ='ML')")
r("偏差値<-round(a[,1],4)*10+50")
#ユーザーの偏差値
value = r.get("偏差値")
data.iat[len(data)-1,16] = value
print(data)
ranking = data.rank(ascending=False,method='max')
print(ranking)
#順位
rank = ranking.iat[len(data)-1,16].astype(int)
#受験人口
every = len(data)
self.render("result.html", value=value,rank=rank ,every=every)
@author: narrowly
"""
import numpy as np
import pandas as pd
import scipy.stats as sct
import random
import pyper
import matplotlib.pyplot as plt
plt.style.use('ggplot')
# 前処理
np.random.seed(4521)
N = 10000 # N_particle
r = pyper.R(use_numpy='True', use_pandas='True')
r("load('data/ArtifitialLocalLevelModel.RData')")
y = r.get('y').astype(np.float64)
t_max = r.get('t_max')
mod = r.get('mod')
# %% コード11-1
# データの整形
y = np.r_[np.nan, y]
# リサンプリング用のインデックス列
k = np.repeat(np.arange(N)[:, np.newaxis], t_max + 1, axis=1)
# 事前分布の設定
# 粒子 (実現値)
x = np.zeros(shape=(t_max + 1, N))
async def plottemp(ctx, self):
r = pyper.R()
r("source(file='/home/rito/Programming/Python/discord_bot/cogs/plot.R')"
)
await self.send("部屋の温湿度と明るさはこんな感じです。")
await self.send(file=discord.File('/home/rito/image/plot.png'))
x[i] = 0.4 * x[i - 1] + 0.8 * x[i - 1] * z[i - 1] + z[i]
halfWindowSize = 20
ret = {}
ret[0] = pyHampel.hampel(x, halfWindowSize, method="center")
#set the center of window at target value
ret[1] = ret[0] # just copy
ret[2] = pyHampel.hampel(x, halfWindowSize, method="same") #same window
ret[3] = pyHampel.hampel(x, halfWindowSize,
method="ignore") # ignore in end
ret[4] = pyHampel.hampel(x, halfWindowSize, method="nan") #set nan in end
print(ret[0][0]) #filtered data
print(ret[0][1]) #indices of outliers
#compare pyhampel with R
r = pyper.R()
r("library(pracma);")
r.assign("x", x)
r("omad <- hampel(x, k=20);")
retR = r.get("omad")
fig, ax = plt.subplots(5, 1, figsize=(8, 8))
ax[0].plot(t, x, "b", label="original data")
for i in range(5):
ax[i].plot(t, retR["y"], "orange", label="R")
ax[i].plot(t, ret[i][0], "r.", label="python")
ax[i].legend(loc="lower center")
ax[0].set_title("overall")
ax[1].set_title("center")
ax[2].set_title("same")
ax[3].set_title("ignore")
data = pd.read_excel("F:\GitHub\FOF\Global Allocation\SBG_US_M.xlsx")
data = pd.read_excel("/Users/WangBin-Mac/FOF/Global Allocation/SBG_US_M.xlsx")
data = data.interpolate()
data = data.dropna().pct_change().dropna()
data
#data_W = data.pct_change().dropna()*100
rp_result_list = []
mu_result_list = []
index_list = []
r = pr.R(use_pandas=True)
r("library(MSwM)")
for each in range(119, len(data) - 1):
#each = 95
#data_M.index[each]
#data_frame = data[:data.index[each]]
data_frame = data[data.index[each - 119]:data.index[each]]
#data_frame = data_frame[['SP500', 'Barclays_US_bond']]
'''
mu_wgt = Ms_MU(data_frame, {'SP500':True, 'Barclays_US_bond':False}, 2)
print each
print mu_wgt
rp_wgt = Ms_RP(data_frame, {'SP500':True, 'Barclays_US_bond':False})
