pavlin-policar

def confidence_histogram(y_true, y_probs, n_bins=10, ax=None):
    if ax is None:
        fig, ax = plt.subplots(figsize=(4, 4))
        
    confidences = np.max(y_probs, axis=1)
    predictions = np.argmax(y_probs, axis=1)
    accuracies = predictions == y_true

    bins = np.linspace(0, 1 - 1 / n_bins, n_bins)
    bin_indices = np.digitize(confidences, bins=bins[1:])

pavlin-policar

import scipy.sparse as sp
import numpy as np


def plot_marker(
    marker,
    dataset,
    embedding: np.ndarray,
    binary=True,
    s=1,

pavlin-policar

from sklearn.base import BaseEstimator
from sklearn.neighbors import NearestNeighbors

import numpy as np
import networkx as nx
from community import best_partition


def jaccard(x: set, y: set) -> float:
    return len(x & y) / len(x | y)

pavlin-policar

from collections import OrderedDict
from typing import Iterable

import numpy as np
import scipy.sparse as sp
import scipy.stats as stats
import pandas as pd


def FDR(p_values: Iterable, dependent=False, m=None, ordered=False) -> Iterable:

pavlin-policar

def _get_required_decimals(x: np.ndarray) -> int:
    """Determine the number of decimals needed to nicely show the data."""
    req_decimals = np.floor(np.log10(x))
    dec = np.percentile(req_decimals, [5])
    return abs(int(dec))
    

pavlin-policar

def smooth_line(x, y, sigma=5, interp_points=300):
    from scipy.ndimage import gaussian_filter1d

    x_interp = np.linspace(x.min(), x.max(), interp_points)

    # The gaussian filter assumes spacing 1, so we have to adjust sigma
    interp_diff = x_interp[1] - x_interp[0]
    effective_sigma = sigma / interp_diff
    
    y_interp = np.interp(x_interp, x, y)