{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# 时间日期和时间增量\n", "\n", "## datetime64 基础\n", "\n", "在 numpy 中，我们很方便的将字符串转换成时间日期类型 `datetime64`（`datetime` 已被 python 包含的日期时间库所占用）。\n", "\n", "`datatime64`是带单位的日期时间类型，其单位如下：\n", "\n", "日期单位 | 代码含义|时间单位 | 代码含义\n", ":---:|:---:|:---:|:---:\n", "Y | 年 |h | 小时\n", "M | 月 |m | 分钟\n", "W | 周 |s | 秒\n", "D | 天 |ms | 毫秒\n", "- | - |us | 微秒\n", "- | - |ns | 纳秒\n", "- | - |ps | 皮秒\n", "- | - |fs | 飞秒\n", "- | - |as | 阿托秒\n", "\n", "注意：\n", "- 1秒 = 1000 毫秒（milliseconds）\n", "- 1毫秒 = 1000 微秒（microseconds）\n", "\n", "【例】从字符串创建 datetime64 类型时，默认情况下，numpy 会根据字符串自动选择对应的单位。\n", "```python\n", "import numpy as np\n", "\n", "a = np.datetime64('2020-03-01')\n", "print(a, a.dtype) # 2020-03-01 datetime64[D]\n", "\n", "a = np.datetime64('2020-03')\n", "print(a, a.dtype) # 2020-03 datetime64[M]\n", "\n", "a = np.datetime64('2020-03-08 20:00:05')\n", "print(a, a.dtype) # 2020-03-08T20:00:05 datetime64[s]\n", "\n", "a = np.datetime64('2020-03-08 20:00')\n", "print(a, a.dtype) # 2020-03-08T20:00 datetime64[m]\n", "\n", "a = np.datetime64('2020-03-08 20')\n", "print(a, a.dtype) # 2020-03-08T20 datetime64[h]\n", "```\n", "\n", "\n", "\n", "\n", "\n", "【例】从字符串创建 datetime64 类型时，可以强制指定使用的单位。\n", "```python\n", "import numpy as np\n", "\n", "a = np.datetime64('2020-03', 'D')\n", "print(a, a.dtype) # 2020-03-01 datetime64[D]\n", "\n", "a = np.datetime64('2020-03', 'Y')\n", "print(a, a.dtype) # 2020 datetime64[Y]\n", "\n", "print(np.datetime64('2020-03') == np.datetime64('2020-03-01')) # True\n", "print(np.datetime64('2020-03') == np.datetime64('2020-03-02')) #False\n", "```\n", "\n", "由上例可以看出，2019-03 和 2019-03-01 所表示的其实是同一个时间。\n", "事实上，如果两个 datetime64 对象具有不同的单位，它们可能仍然代表相同的时刻。并且从较大的单位（如月份）转换为较小的单位（如天数）是安全的。\n", "\n", "\n", "【例】从字符串创建 datetime64 数组时，如果单位不统一，则一律转化成其中最小的单位。\n", "```python\n", "import numpy as np\n", "\n", "a = np.array(['2020-03', '2020-03-08', '2020-03-08 20:00'], dtype='datetime64')\n", "print(a, a.dtype)\n", "# ['2020-03-01T00:00' '2020-03-08T00:00' '2020-03-08T20:00'] datetime64[m]\n", "```\n", "\n", "\n", "【例】使用`arange()`创建 datetime64 数组，用于生成日期范围。\n", "```python\n", "import numpy as np\n", "\n", "a = np.arange('2020-08-01', '2020-08-10', dtype=np.datetime64)\n", "print(a)\n", "# ['2020-08-01' '2020-08-02' '2020-08-03' '2020-08-04' '2020-08-05'\n", "# '2020-08-06' '2020-08-07' '2020-08-08' '2020-08-09']\n", "print(a.dtype) # datetime64[D]\n", "\n", "a = np.arange('2020-08-01 20:00', '2020-08-10', dtype=np.datetime64)\n", "print(a)\n", "# ['2020-08-01T20:00' '2020-08-01T20:01' '2020-08-01T20:02' ...\n", "# '2020-08-09T23:57' '2020-08-09T23:58' '2020-08-09T23:59']\n", "print(a.dtype) # datetime64[m]\n", "\n", "a = np.arange('2020-05', '2020-12', dtype=np.datetime64)\n", "print(a)\n", "# ['2020-05' '2020-06' '2020-07' '2020-08' '2020-09' '2020-10' '2020-11']\n", "print(a.dtype) # datetime64[M]\n", "```\n", "\n", "## datetime64 和 timedelta64 运算\n", "\n", "【例】timedelta64 表示两个 datetime64 之间的差。timedelta64 也是带单位的，并且和相减运算中的两个 datetime64 中的较小的单位保持一致。\n", "```python\n", "import numpy as np\n", "\n", "a = np.datetime64('2020-03-08') - np.datetime64('2020-03-07')\n", "b = np.datetime64('2020-03-08') - np.datetime64('202-03-07 08:00')\n", "c = np.datetime64('2020-03-08') - np.datetime64('2020-03-07 23:00', 'D')\n", "\n", "print(a, a.dtype) # 1 days timedelta64[D]\n", "print(b, b.dtype) # 956178240 minutes timedelta64[m]\n", "print(c, c.dtype) # 1 days timedelta64[D]\n", "\n", "a = np.datetime64('2020-03') + np.timedelta64(20, 'D')\n", "b = np.datetime64('2020-06-15 00:00') + np.timedelta64(12, 'h')\n", "print(a, a.dtype) # 2020-03-21 datetime64[D]\n", "print(b, b.dtype) # 2020-06-15T12:00 datetime64[m]\n", "```\n", "\n", "【例】生成 timedelta64时，要注意年（'Y'）和月（'M'）这两个单位无法和其它单位进行运算（一年有几天？一个月有几个小时？这些都是不确定的）。\n", "```python\n", "import numpy as np\n", "\n", "a = np.timedelta64(1, 'Y')\n", "b = np.timedelta64(a, 'M')\n", "print(a) # 1 years\n", "print(b) # 12 months\n", "\n", "c = np.timedelta64(1, 'h')\n", "d = np.timedelta64(c, 'm')\n", "print(c) # 1 hours\n", "print(d) # 60 minutes\n", "\n", "print(np.timedelta64(a, 'D'))\n", "# TypeError: Cannot cast NumPy timedelta64 scalar from metadata [Y] to [D] according to the rule 'same_kind'\n", "\n", "print(np.timedelta64(b, 'D'))\n", "# TypeError: Cannot cast NumPy timedelta64 scalar from metadata [M] to [D] according to the rule 'same_kind'\n", "```\n", "\n", "\n", "【例】timedelta64 的运算。\n", "```python\n", "import numpy as np\n", "\n", "a = np.timedelta64(1, 'Y')\n", "b = np.timedelta64(6, 'M')\n", "c = np.timedelta64(1, 'W')\n", "d = np.timedelta64(1, 'D')\n", "e = np.timedelta64(10, 'D')\n", "\n", "print(a) # 1 years\n", "print(b) # 6 months\n", "print(a + b) # 18 months\n", "print(a - b) # 6 months\n", "print(2 * a) # 2 years\n", "print(a / b) # 2.0\n", "print(c / d) # 7.0\n", "print(c % e) # 7 days\n", "```\n", "\n", "【例】numpy.datetime64 与 datetime.datetime 相互转换\n", "```python\n", "import numpy as np\n", "import datetime\n", "\n", "dt = datetime.datetime(year=2020, month=6, day=1, hour=20, minute=5, second=30)\n", "dt64 = np.datetime64(dt, 's')\n", "print(dt64, dt64.dtype)\n", "# 2020-06-01T20:05:30 datetime64[s]\n", "\n", "dt2 = dt64.astype(datetime.datetime)\n", "print(dt2, type(dt2))\n", "# 2020-06-01 20:05:30 \n", "```\n", "\n", "## datetime64 的应用\n", "\n", "为了允许在只有一周中某些日子有效的上下文中使用日期时间，NumPy包含一组“busday”（工作日）功能。\n", "\n", "- `numpy.busday_offset(dates, offsets, roll='raise', weekmask='1111100', holidays=None, busdaycal=None, out=None)` First adjusts the date to fall on a valid day according to the roll rule, then applies offsets to the given dates counted in valid days.\n", "\n", "参数`roll`：{'raise', 'nat', 'forward', 'following', 'backward', 'preceding', 'modifiedfollowing', 'modifiedpreceding'}\n", "- 'raise' means to raise an exception for an invalid day.\n", "- 'nat' means to return a NaT (not-a-time) for an invalid day.\n", "- 'forward' and 'following' mean to take the first valid day later in time.\n", "- 'backward' and 'preceding' mean to take the first valid day earlier in time.\n", "\n", "\n", "【例】将指定的偏移量应用于工作日，单位天（'D'）。计算下一个工作日，如果当前日期为非工作日，默认报错。可以指定 `forward` 或 `backward` 规则来避免报错。（一个是向前取第一个有效的工作日，一个是向后取第一个有效的工作日）\n", "```python\n", "import numpy as np\n", "\n", "# 2020-07-10 星期五\n", "a = np.busday_offset('2020-07-10', offsets=1)\n", "print(a) # 2020-07-13\n", "\n", "a = np.busday_offset('2020-07-11', offsets=1)\n", "print(a)\n", "# ValueError: Non-business day date in busday_offset\n", "\n", "a = np.busday_offset('2020-07-11', offsets=0, roll='forward')\n", "b = np.busday_offset('2020-07-11', offsets=0, roll='backward')\n", "print(a) # 2020-07-13\n", "print(b) # 2020-07-10\n", "\n", "a = np.busday_offset('2020-07-11', offsets=1, roll='forward')\n", "b = np.busday_offset('2020-07-11', offsets=1, roll='backward')\n", "print(a) # 2020-07-14\n", "print(b) # 2020-07-13\n", "```\n", "\n", "可以指定偏移量为 0 来获取当前日期向前或向后最近的工作日，当然，如果当前日期本身就是工作日，则直接返回当前日期。\n", "\n", "- `numpy.is_busday(dates, weekmask='1111100', holidays=None, busdaycal=None, out=None)` Calculates which of the given dates are valid days, and which are not.\n", "\n", "【例】返回指定日期是否是工作日。\n", "```python\n", "import numpy as np\n", "\n", "# 2020-07-10 星期五\n", "a = np.is_busday('2020-07-10')\n", "b = np.is_busday('2020-07-11')\n", "print(a) # True\n", "print(b) # False\n", "```\n", "\n", "【例】统计一个 `datetime64[D]` 数组中的工作日天数。\n", "```python\n", "import numpy as np\n", "\n", "# 2020-07-10 星期五\n", "begindates = np.datetime64('2020-07-10')\n", "enddates = np.datetime64('2020-07-20')\n", "a = np.arange(begindates, enddates, dtype='datetime64')\n", "b = np.count_nonzero(np.is_busday(a))\n", "print(a)\n", "# ['2020-07-10' '2020-07-11' '2020-07-12' '2020-07-13' '2020-07-14'\n", "# '2020-07-15' '2020-07-16' '2020-07-17' '2020-07-18' '2020-07-19']\n", "print(b) # 6\n", "```\n", "\n", "\n", "【例】自定义周掩码值，即指定一周中哪些星期是工作日。\n", "```python\n", "import numpy as np\n", "\n", "# 2020-07-10 星期五\n", "a = np.is_busday('2020-07-10', weekmask=[1, 1, 1, 1, 1, 0, 0])\n", "b = np.is_busday('2020-07-10', weekmask=[1, 1, 1, 1, 0, 0, 1])\n", "print(a) # True\n", "print(b) # False\n", "```\n", "\n", "- `numpy.busday_count(begindates, enddates, weekmask='1111100', holidays=[], busdaycal=None, out=None)`Counts the number of valid days between `begindates` and `enddates`, not including the day of `enddates`.\n", "\n", "【例】返回两个日期之间的工作日数量。\n", "```python\n", "import numpy as np\n", "\n", "# 2020-07-10 星期五\n", "begindates = np.datetime64('2020-07-10')\n", "enddates = np.datetime64('2020-07-20')\n", "a = np.busday_count(begindates, enddates)\n", "b = np.busday_count(enddates, begindates)\n", "print(a) # 6\n", "print(b) # -6\n", "```\n", "\n", "\n", "---\n", "参考图文\n", "\n", "- https://www.jianshu.com/p/336cd77d9914\n", "- https://www.cnblogs.com/gl1573/p/10549547.html#h2datetime64\n", "- https://www.numpy.org.cn/reference/arrays/datetime.html#%E6%97%A5%E6%9C%9F%E6%97%B6%E9%97%B4%E5%8D%95%E4%BD%8D\n", "\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.8.3" }, "toc": { "base_numbering": 1, "nav_menu": {}, "number_sections": true, "sideBar": true, "skip_h1_title": false, "title_cell": "Table of Contents", "title_sidebar": "Contents", "toc_cell": false, "toc_position": {}, "toc_section_display": true, "toc_window_display": true }, "varInspector": { "cols": { "lenName": 16, "lenType": 16, "lenVar": 40 }, "kernels_config": { "python": { "delete_cmd_postfix": "", "delete_cmd_prefix": "del ", "library": "var_list.py", "varRefreshCmd": "print(var_dic_list())" }, "r": { "delete_cmd_postfix": ") ", "delete_cmd_prefix": "rm(", "library": "var_list.r", "varRefreshCmd": "cat(var_dic_list()) " } }, "types_to_exclude": [ "module", "function", "builtin_function_or_method", "instance", "_Feature" ], "window_display": false } }, "nbformat": 4, "nbformat_minor": 4 }