Jan 01, 1970
{ '09/09/2023': [13, 17, 24, 30, 35, 37], '07/09/2023': [7, 17, 19, 25, 35, 37], '05/09/2023': [2, 3, 5, 9, 36, 37], '02/09/2023': [4, 12, 22, 27, 30, 34], '29/08/2023': [6, 8, 15, 19, 26, 31], '26/08/2023': [6, 7, 14, 21, 25, 34], '22/08/2023': [2, 6, 10, 23, 24, 29], ... }
numbers = np.array(list(lotto.values())).flatten() [13, 17, 24, 30, 35, 37, 7, 17, 19, 25, 35, 37, 2, 3, 5, 9, 36, ...]
count = np.bincount(numbers)[1:] [268, 256, 257, 242, 255, 273, 247, 277, 260, 267, 289, 294, 271, 239, 254, 255, 263, 243, 246, 271, 265, 254, 252, 243, 291, 271, 258, 264, 275, 258, 251, 244, 263, 256, 267, 251, 264]
দেখা যাচ্ছে লটারির ফলাফলে সংখ্যার বণ্টন তুলনামূলকভাবে সমান। এটি আরও নিশ্চিত করার জন্য, আমরা বিতরণের সমানতা যাচাই করার জন্য একটি পরীক্ষা পরিচালনা করতে পারি।
def chi2(data, size, expect, p_value = 0.05): pl = size * 1/expect df = expect - 1 x2_crit_1 = stats.chi2.ppf(p_value, df) x2_crit_2 = stats.chi2.ppf(1 - p_value, df) x2 = 0 for i in range(expect): x2 += ((data[i] - pl) ** 2)/pl accepted = x2_crit_1 < x2 < x2_crit_2 if x2_crit_1 < x2_crit_2 else x2_crit_2 < x2 < x2_crit_1 return x2, accepted
এই ফাংশনটি chi-squared পরিসংখ্যান সমন্বিত টিপল প্রদান করে এবং এর ফলে সম্ভাব্যতা 1 - 2 * p-value
, অর্থাৎ, এই বিচ্ছিন্ন ইউনিফর্ম ডিস্ট্রিবিউশনের চরম মানগুলির সম্ভাবনা কম থাকে।
N = 37 chi2(count, len(numbers), N) (25.0748, True)
from scipy import stats chi2_statistic, p_value = stats.chisquare(count) (25.074, 0.96053)
from itertools import combinations pairs = list(combinations(range(1, N), 2))
pairs_count = np.zeros([N] * 2, dtype=int) for pair in pairs: for draw in lotto.values(): if pair[0] in draw and pair[1] in draw: pairs_count[pair[0]][pair[1]] += 1 pairs_count = pairs_count[1:, 1:]
counts = pairs_count.flatten() counts = counts[counts > 0] chi2(counts, sum(counts), len(counts)) (589.2721893491138, True)
chi2_statistic, p_value = stats.chisquare(counts) (589.2721893491124, 0.8698507423203673)
comb3 = list(combinations(range(1, N), 3)) comb3_count = np.zeros([N] * 3, dtype=int) for comb in comb3: for draw in lotto.values(): contains = comb[0] in draw and comb[1] in draw and comb[2] in draw if contains: comb3_count[comb[0]][comb[1]][comb[2]] += 1 comb3_count = comb3_count[1:, 1:, 1:] counts = comb3_count.flatten() counts = counts[counts > 0] chi2(counts, sum(counts), len(counts)) (6457.575829383709, False)
6457.575829383709 < 6840.049842653838
chi2_statistic, p_value = stats.chisquare(counts) (6457.575829383886, 0.9999997038479482)
count.argmax() or list(count).index(max(count)) 11
year_result = dict() for year in range(2009, 2024): new_dict = {k:v for (k,v) in lotto.items() if str(year) in k} year_result[year] = np.bincount(np.array(list(new_dict.values())).flatten())[1:].argmax() { 2009: 16, 2010: 10, 2011: 11, 2012: 24, 2013: 32, 2014: 34, 2015: 21, 2016: 25, 2017: 5, 2018: 10, 2019: 24, 2020: 11, 2021: 12, 2022: 14, 2023: 11 }
year_result = dict() arr = [] for year in range(2009, 2024): new_dict = {k:v for (k,v) in lotto.items() if str(year) in k} arr += list(np.array(list(new_dict.values())).flatten()) year_result['2009 - ' + str(year) if year > 2009 else str(year)] = np.bincount(arr)[1:].argmax() { '2009': 16, '2009 - 2010': 10, '2009 - 2011': 11, '2009 - 2012': 20, '2009 - 2013': 20, '2009 - 2014': 20, '2009 - 2015': 34, '2009 - 2016': 20, '2009 - 2017': 10, '2009 - 2018': 10, '2009 - 2019': 10, '2009 - 2020': 10, '2009 - 2021': 10, '2009 - 2022': 24, '2009 - 2023': 11 }
lotto_counts = {} for k, v in lotto.items(): v_str = str(v) if v_str in lotto_counts: lotto_counts[v_str] += [k] else: lotto_counts[v_str] = [k] result = {k: v for k, v in lotto_counts.items() if len(lotto_counts[k]) > 1} { '[13, 14, 26, 32, 33, 36]': ['16/10/2010', '21/09/2010'] }